A systematic review on frailty indices utilized in oncologic surgery and radiotherapy for metastatic spine disease: A critical appraisal of components and performance

Highlights

• •Frailty indices for metastatic spine disease show inconsistent outcome prediction.
• •None of the frailty indices captures patient energy measures.
• •A novel frailty index is needed to guide treatment between surgery and radiotherapy.

Abstract

Background

Frailty assessment is becoming increasingly important for risk stratification in metastatic spine disease (MSD) management. However, the optimal frailty assessment tool for this population remains undefined. The aim of this systematic review was to critically evaluate frailty indices utilized in surgical and radiotherapeutic management of MSD by appraising their components and predictive performance.

Methods

We systematically searched PubMed, Cochrane, and Epistemonikos from inception until November 13, 2024 for studies exploring frailty indices in MSD management. Methodological quality assessment was undertaken using the Methodological index for nonrandomized studies (MINORS) and Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tools. We evaluated indices for comprehensiveness by assessing if they included measures across 8 dimensions (comorbidity burden, mobility, cognition, mood, social vulnerability, nutrition, energy, and function). We also assessed their predictive utility for mortality and complications.

Results

Out of a total of 42 studies, 15 studies reporting on 61,663 patients (Mean age: 63 years; 44% female) met the inclusion criteria. All studies were of moderate quality based on MINORS analysis. Six frailty indices were identified, with all including comorbidities while none incorporating energy measures. The Hospital Frailty Risk Score was the most comprehensive (75% of frailty spectrum). Predictive performance varied considerably across studies, with inconsistent associations with complications and mortality after both: surgical and radiotherapeutic interventions.

Conclusions

Contemporary frailty indices showed substantial heterogeneity in both composition and predictive performance for MSD outcomes. The inconsistent performance and incomplete capture of frailty dimensions underscored the need for developing a novel frailty index that incorporates oncologic factors and balances comprehensive evaluation with clinical feasibility to help guide treatment decisions between surgery and radiotherapy for patients with MSD.

Introduction

Frailty is a complex, multidimensional syndrome characterized by decreased physiological reserve and increased vulnerability to stressors [1]. It is increasingly recognized as a critical factor in predicting health outcomes, particularly in older adults and those with chronic diseases. The prevalence of frailty increases with age, affecting approximately 20% of community-dwelling adults over 65 years of age [2]. For patients with metastatic spine disease (MSD), the prevalence has been estimated to be as high as 72% [3].

The conceptual framework popularized by Rockwood and Mitnitski defines frailty as a cumulative deficit model, encompassing various domains including age, comorbidities, functional status, cognition, and physiological measures [4]. This comprehensive approach underscores frailty as more than just a collection of comorbidities, but rather a holistic representation of an individual's overall health status. The application of frailty assessment in spinal oncology presents unique challenges compared to other surgical populations. A recent systematic review by MacLean MA et al. evaluated frailty and sarcopenia tools specifically in the spinal oncology context, finding significant limitations in their clinimetric properties [5].

Spinal metastases are a significant complication of cancer, affecting from 10% to 15% of patients during their disease course [6,7]. In the context of MSD, frailty assessment has emerged as a potentially valuable tool for risk stratification and outcome prediction [8,9]. However, while various frailty indices have been applied in this setting, significant gaps persist in our understanding. Specifically, the overall performance of these indices in predicting MSD outcomes after surgical and / or radiotherapeutic management has not been systematically evaluated. Moreover, it is unclear whether the frailty measures currently in use adequately represent the multidimensional, cumulative deficit model of frailty as conceptualized by Rockwood and Mitnitski [4].

Given these significant knowledge gaps, this systematic review was undertaken to obtain a holistic overview of frailty indices being adopted to predict MSD outcomes after surgical and / or radiotherapeutic management. By critically appraising the components and performance of various frailty indices, this review sought to provide insights that could guide the selection and application of the most appropriate frailty assessment tools in this challenging patient population.

Materials and methods

Search strategy

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines, [10]. following a prospective study protocol. A literature search was performed through PubMed/Medline, Cochrane Database of Systematic Reviews (CDSR), and Epistemonikos from inception to November 13, 2024. The search strategy included the terms "spin* metastas*" [Title/Abstract] and "Frailty" [MeSH Terms] (detailed in the Supplementary Materials). Reference lists of eligible studies were also searched for relevant publications. Three researchers [REDACTED FOR ANONYMITY] independently reviewed articles, extracted data, and assessed methodological quality.

Objectives and endpoints

Our objectives were: (1) To explore which frailty indices have been explored in the context of MSD management with surgery and / or radiotherapy; (2) To determine the components that are part of the indices; and (3) To assess the general performance of these indices in MSD outcome prediction. Although there is no universal consensus on the components a frailty index should have, it is agreed by most that it is a multidimensional condition. Therefore, [RDDACTED FOR ANONYMITY] independently assessed whether the utilized frailty indices included measures of disease burden (comorbidities), mobility, cognition, mood, social vulnerability, nutrition, energy, and/or function [2,11].

These 8 dimensions were selected based on established conceptual frameworks of frailty, particularly the cumulative deficit model by Rockwood and Mitnitski [4]. Each dimension represents a distinct aspect of frailty: (1) Disease burden / comorbidities (Presence of chronic conditions such as cardiovascular disease or diabetes mellitus); (2) Mobility (Measures of gait speed, balance, or use of mobility aids); (3) Cognition (Assessment of cognitive function, memory, or executive function); (4) Mood (Evaluation of depression, anxiety, psychological well-being); (5) Social vulnerability (Social support, living situation, or caregiver availability); (6) Nutrition (Nutritional status, weight loss, BMI, or albumin levels); (7) Energy (Measures of fatigue, exhaustion, or activity levels); and (8) Function (Activities of daily living or performance status). By evaluating frailty indices against these 8 dimensions, we could assess how comprehensively each index captured the multidimensional nature of frailty as conceptualized in contemporary geriatric literature. An index was considered to include a dimension if it contained at least 1 variable measuring that aspect of frailty.

Study selection

Studies were included if they met the following criteria: (1) Metastatic spine tumor patients as the primary population; (2) Patients treated with surgery and / or radiation; (3) Frailty indices used to predict mortality or complications; (4) Univariable and / or multivariable analyses reported for frailty indices and outcomes of interest; and (5) Retrospective or prospective cohort studies, or randomized-controlled trials published in English language. The study selection process followed the updated PRISMA flow diagram.

Data extraction included author identifiers, study characteristics, patient demographics, clinical characteristics, tumor details, treatment information, frailty indices used with their cut-offs, and outcome analyses.

Quality assessment and risk of bias

The Methodological index for nonrandomized studies (MINORS) tool was used to assess methodological quality of included studies, [12] with global ideal scores of 16 for noncomparative and 24 for comparative studies [13]. For machine learning studies, the Quality Assessment of Diagnostic Accuracy Studies—version 2 (QUADAS-2) tool was applied to assess risk of bias across 4 domains: patient selection, index test, reference standard, and flow and timing [14].

Statistical analysis

Exploratory data analysis was performed with categorical variables expressed as percentages and continuous variables as mean with standard deviation.

Results

A search carried out through PubMed/Medline, CDSR, and Epistemonikos databases yielded 44 articles. After removing 4 duplicates, the abstracts and, where required, full-texts of 40 studies were considered in accordance with the aforementioned inclusion eligibility criteria. Two studies were included on manual exploration. Fifteen studies satisfied the inclusion eligibility criteria and were included in the final review [3,[15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]] (Table 1). The underlying causes of the exclusion are concisely indicated in the relevant component of the PRISMA flow diagram. (Fig. 1) The complete list of excluded studies is shared in the “Supplementary Materials.”

Table 1.

Demographics of studies exploring frailty indices for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

Author	Data source and treatment	Total patients (male and female)	Age	Location of primary disease	Frailty index	Studied Outcome	Results
Bang C et al. [15]	Single-institution database Surgery	69 patients (75 cases) 28 male (32%) and 47 female (68%)	Mean: 61.8 (26 – 83) years	-Lung: 30.66% (n=23 cases) -Prostate: 12% (n=9 cases) -Breast: 10.66% (n=8 cases) -Kidney: 9.33% (n=7 cases) -Myeloma: 8% (n=6 cases) -Colorectal: 5.33% (n=4 cases) -Liver: 4% (n=3 cases) -Hepatobiliary: 2.66% (2 cases) -Gastric: 2.66% (2 cases) -Lymphoma: 2.66% (2 cases) -Thyroid: 1.33% (n=1 case) -Others: 10.66% (n=8 cases)	Modified 5-Item Frailty Index (mFI-5)	Perioperative medical complications and postoperative ambulatory status	mFI-5 could not significantly predict perioperative medical complications and postoperative ambulatory status after surgery for metastatic spine disease (MSD).
Elsamadicy AA et al. [3]	American College of Surgeons – National Surgical Quality Improvement Program (ACS NSQIP) database Surgery	1796 patients 1116 male (62%) and 680 female (38%)	Mean: 60.95 years	Not reported	Revised-Risk Analysis Index (RAI-rev) and mFI-5	Length of hospital stay, 30-day readmission, 30-day complications and 30-day mortality	Both: RAI-rev and mFI-5 independently predicted length of hospital stay and 30-day readmission. In contrast to mFI-5, RAI-rev independently predicted 30-day complications and 30-day mortality. RAI-rev was, therefore, more robust in predicting outcomes after surgery for MSD.
Elsamadicy AA et al., [16]	Nationwide Readmission Database Surgery	4346 patients 2653 male (61%) and 1693 female (39%)	Mean: 64.33 years	Not reported	Hospital Frailty Risk Score (HFRS)	30-day readmission	HFRS tertiles independently predicted 30-day readmission after surgery for MSD.
Musharbash FN et al., [17]	ACS NSQIP database Surgery	1195 patients 782 male (65%) and 413 female (35%)	Mean: 61 (±12) years	Not reported	mFI-5	30-day mortality	Amongst all the predictors explored, mFI-5 most poorly prognosticated 30-day mortality after surgery for MSD.
Wu LC et al., [18]	National Inpatient Sample (NIS) database Surgery	31925 patients 16104 male (50%) and 15901 female (50%)	Mean: 63.5 (±0.1) years	-Lung: 13.61% (n=4345 patients) -Breast: 9.31% (n=2974 patients) -Prostate: 8.71% (n=2783 patients) -Hematologic: 7.59% (n=2426 patients) -Kidney: 5.27% (n=1683 patients) -Colorectal: 2.29% (n=732 patients) -Liver: 1.26% (n=404 patients) -Others / unspecified: 51.92% (n=16578)	Metastatic Spinal Tumor Frailty Index (MSTFI)	Life-threatening complications	Irrespective of the age of the patient, MSTFI independently predicted life-threatening complications after surgery for MSD.
Elsamadicy AA et al., [19]	NIS database Surgery	11480 patients 6999 male (61%) and 4481 female (39%)	Mean: 65.03 years	Not reported	HFRS	Length of hospital stay, discharge disposition, and total hospital costs	HFRS tertiles independently predicted increased length of hospital stay, nonroutine discharge disposition, and increased total hospital cost after surgery for MSD.
Massaad E et al., [20]	Single-institution database Surgery	303 patients 175 male (58%) and 128 female (42%)	Mean: 62 (±11.91) years	-Genitourinary: 25.41% (n=77 patients) -Lung: 19.8% (n=60 patients) -Gastrointestinal tract: 14.19% (n=43 patients) -Breast: 10.89% (n=33 patients) -Hematology: 7.26% (n=22 patients) -Skin: 4.95% (n=15 patients) -Bone sarcoma: 4.29% (n=13 patients) -Head & neck: 3.63% (n=11 patients) -Thyroid: 3.3% (n=10 patients) -Soft-tissue sarcoma: 2.97% (n=9 patients) -Other: 2.97% (n=9 patients)	MSTFI	All-cause mortality	MSTFI quartiles independently predicted all-cause mortality after surgery for MSD.
Bakhsheshian J et al. [21]	Healthcare Cost and Utilization Project Nationwide Database Surgery	3949 patients 2510 male (64%) and 1439 female (36%)	Mean: 64.65 years	-Prostate: 16.2% (n=640 patients) -Lung: 14.05% (n=555 patients) -Kidney: 7.67% (n=303 patients) -Breast: 7.52% (n=297 patients) -Gastrointestinal / Digestive Organs: 4.58% (n=181 patients) -Lymphoma / Myeloma: 2.43% (n=96 patients) -Melanoma: 2.02% (n=80 patients) -Thyroid: 1.84% (n=73 patients) -Sarcoma: 1.11% (n=44 patients) -Unspecified: 42.54% (n=1680 patients)	Johns Hopkins Adjusted Clinical Groups – Frailty Index (JHACG-FI)	Medical complications, neurological complications, surgical complications, length of hospital stay, discharge disposition, hospitalization cost, and mortality	JHACG-FI independently predicted medical and surgical complications as well as length of hospital stay, nonroutine discharge and hospitalization cost across all spine regions and crossing junctions. However, although JHACG-FI was reported to independently predict mortality in lumbar crossing junctions segments, it could not predict mortality after surgery for MSD in cervical and thoracic regions. Moreover, it could not predict neurological complications, either.
Hersh AM et al., [22]	Single-institution database Surgery	322 patients 197 male (61%) and 125 female (39%)	Mean: 59.5 (±12) years	Complete breakdown not reported	mFI-5 and MSTFI	Length of hospital stay, occurrence of ≥ 1 postoperative complications, and discharge disposition	Both: mFI-5 and MSTFI robustly prognosticated length of hospital stay, occurrence of ≥ 1 postoperative complications, and nonroutine discharge after surgery for MSD.
Massaad E et al., [23]	Database from 2 institutions Surgery and adjuvant radiotherapy	88 patients 65 male (74%) and 23 female (26%)	Mean: 60.8 (±11.4) years	-Kidney: 100% (n=88 patients)	Modified 11-Item Frailty Index (mFI-11)	Overall survival	mFI-11 tertiles were not significantly associated with overall survival of patients with primary renal cell carcinoma managed with surgical intervention and adjuvant radiotherapy.
Massaad E et al., [24]	Single-institution database Surgery	479 patients 281 male (59%) and 198 female (41%)	Mean: 62.4 (±12.1) years	-Lung: 18.78% (n=90 patients) -Kidney: 10.64% (n=51 patients) -Breast: 9.81% (n=47 patients) -Blood (heme): 9.18% (n=44 patients) -Prostate: 8.76% (n=42 patients) -Colon / rectal: 5.84% (n=28 patients) -Bone sarcoma: 5.01% (n=24 patients) -Thyroid: 4.8% (n=23 patients) -Head & neck: 4.59% (n=22 patients) -Skin: 4.38% (n=21 patients) -Other: 17.95% (n=86 patients)	MSTFI	Postoperative complications, length of hospital stay, and in-hospital mortality	MSTFI tertiles were not significantly associated with postoperative complications, length of hospital stay, and in-hospital mortality after surgery for MSD.
Zakaria HM et al., [25]	Single-institution database Radiotherapy	369 patients 182 male (49%) and 187 female (51%)	Mean: 65.3 (±11.5) years	-Lung: 44% (n=163 patients) -Breast: 26% (n=96 patients) -Prostate: 20% (n=75 patients) -Myeloma: 9% (n=35 patients)	mFI-5	Overall survival	mFI-5 quartiles were not significantly associated with overall survival of patients with MSD managed with stereotactic body radiation therapy.
Bourassa-Moreau É et al., [26]	Single-institution database Surgery	108 patients 57 male (53%) and 51 female (47%)	Mean: 61.5 (±11.5) years	-Breast: 22.2% (n=24 patients) -Lung: 16.6% (n=18 patients) -Kidney: 15.74% (n=17 patients) -Prostate: 12.03% (n=13 patients) -Other: 33.3% (n=36 patients)	mFI-11 and MSTFI	Adverse events, 1-month and 3-months mortality	Neither mFI-11 nor MSTFI could prognosticate adverse events or mortality significantly after emergent surgery for MSD.
Mohd Rothi I et al., [27]	Single-institution database Surgery	41 patients 26 male (63%) and 15 female (37%)	Mean: 64 (±9.1) years	-Prostate: 26.82% (n=11 patients) -Breast: 14.63% (n=6 patients) -Lungs: 12.19% (n=5 patients) -Renal: 9.75% (n=4 patients) -Bladder: 4.87% (n=2 patients) -Cervix: 2.43% (n=1 patient) -Thyroid: 2.43% (n=1 patient) -Unknown: 14.63% (n=6 patients)	mFI-11	Survival	mFI-11 was not significantly associated with survival after surgery for MSD.
De la Garza Ramos R et al., [28]	NIS database* + ACS-NSQIP database Surgery	4583 patients* + 610 patients 2650 male (58%) and 1931 female (42%)	Median: 62 (54 – 70) years	-Lung: 34.1% (n=1563 patients) -Breast: 21.09% (n=967 patients) -Prostate: 21.09% (n=967 patients) -Kidney: 19.89% (n=912 patients) -Thyroid: 3.79% (n=174 patients)	MSTFI	Major complications, length of hospital stay, and in-hospital mortality	In the internal validation, MSTFI tertiles were significantly associated with major complications, length of hospital stay, and in-hospital mortality. Upon external validation, MSTFI quartiles was not significantly associated with in-hospital mortality, whereas they were significantly associated with major complications.

*The breakdown of female gender, and location of primary disease were reported only for the NIS cohort.

^†The measure of effect size was not reported.

mFI-5, Modified 5-Item Frailty Index; MSD, Metastatic spine disease; ACS NSQIP, American College of Surgeons – National Surgical Quality Improvement Program; RAI-rev, Revised-Risk Analysis Index; HFRS, Hospital Frailty Risk Score; NIS, National Inpatient Sample; MSTFI, Metastatic Spinal Tumor Frailty Index; JHACG-FI, Johns Hopkins Adjusted Clinical Groups – Frailty Index; mFI-11, Modified 11-Item Frailty Index.

Fig. 1.

PRISMA flow-chart illustrating search, selection, and inclusion of studies exploring frailty indices for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

General patient demographics and studies characteristics

The included fifteen studies were published between 2016 and 2024 and reported on a total of 61, 663 patients with the mean age of 63 years and 44% (n=27, 312) females. All included studies were retrospective in design, with an evidence level of Level II: Prognostic Study. Only 40% (n=6) studies reported patient race, [3,[18], [19], [20],22,25] whereas 33% (n=5) studies reported ethnicity of the patients [3,[18], [19], [20],25]. Moreover, only 27% studies (n=4) reported the insurance status of the included patients, [16,18,19,21] and only 20% (n=3) studies reported household income [16,18,19]. None of the studies reported education level and employment status of the included patients. The presence of caregiver support was also not reported.

The studies reported multiple comorbidities, including cardiac disease, diabetes mellitus, and hypertension (Supplementary table 1). ASA [3,22] and CCI [18,22] were adopted by 13% (n=2) each, whereas ECI was adopted by 7% (n=1) studies [21] (Supplementary Table 1). None of the studies reported spinal instability scores. Albumin deficiency was reported by 7% (n=1) studies, [17] and BMI was reported by 27% (n=4) studies [3,17,20,22]. 20% (n=3) studies reported malnutrition (including protein-calorie malnutrition) with an undefined scale, [16,24,28] and the Prognostic Nutritional Index was reported by 7% (n=1) studies only [23] (Supplementary Table 1).

Location of primary tumor was reported by 67% (n=10) studies, with lung cancer recognized as the most common primary tumor overall [15,18,20,21,[23], [24], [25], [26], [27], [28]] (Table 1). Four studies (27%) reported concurrent nonspinal metastases including concomitant node involvement, [21,23,26,28] 33% (n=5) studies reported the surgical approach, [21,22,24,26,28] and 67% (n=10) studies reported the surgical intervention type [3,[16], [17], [18], [19],21,22,26,28] (Supplementary Table 1). Posterior fusion was found to be the most common surgical intervention and approach.

Two studies (13%) reported radiotherapeutic management [23,25]. Adjuvant radiotherapy (either stereotactic radiosurgery or fractionated external beam radiation therapy [EBRT]) was provided to all patients for locoregional control in the Massaad E et al. [23] study. In the Zakaria HM et al. study, stereotactic body radiation therapy (SBRT) was adopted as the primary treatment methodology [25].

All studies were of moderate quality based on MINORS analysis (Supplementary Tables 2 and 3). None of the studies prospectively calculated the study size, or explicitly reported inclusion of consecutive patients. However, all included studies had end-points and follow-up periods appropriate to the respective aim of the studies (Supplementary Tables 2 and 3). Furthermore, all 3 studies that adopted ML had a low risk of bias in the 'Flow and Timing' domain, whereas they had an unclear risk of bias across both: Index test (ML) and reference standard (frailty index) domains, primarily since it was unclear whether each of these were implemented independently, without knowledge of the results of the other [16,20,24] (Fig. 2).

Fig. 2.

The Quality Assessment of Diagnostic Accuracy Studies—version 2 (QUADAS-2) analysis of the included studies exploring frailty indices for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

Frailty indices

Six frailty indices have been explored for prediction of postinterventional complications and mortality after MSD management with surgery and / or radiotherapy (Table 2). The metastatic spinal tumor frailty index (MSTFI) was developed and externally validated in 2016 using the National Inpatient Sample (NIS) and ACS-NSQIP databases [28]. The index was initially developed to identify predictive factors of major postoperative complications after surgery for metastatic spinal tumors and assessed 25 different variables (comorbidities, nutrition, type of admission, and operative approach) identified by diagnostic codes (ICD-9). Following multivariable logistic regression analysis, 9 variables were found to remain significant and thus were included in the index [28]. Furthermore, the MSTFI was also found to be predictive of inpatient mortality and length of stay. In the original investigation, it achieved an overall area under curve (AUC) of 0.67 for complications and 0.56 for mortality [28]. However, it was stratified by tumor type, achieving higher discriminatory ability for patients with prostate and thyroid cancer especially.

Table 2.

Individual components, weights, and methods of frailty stratification of frailty indices adopted for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

Frailty index	Source of development	Outcomes targeted for original development	Individual components with weights for frailty measurement	Method of frailty stratification
Metastatic Spinal Tumor Frailty Index (MSTFI)	National Inpatient Sample database and American College of Surgeons – National Surgical Quality Improvement Program (ACS NSQIP) database with variables identified by diagnostic codes from International Classification of Diseases (ICD), Ninth Revision (ICD-9)	Major complications, length of hospital stay, and in-hospital mortality after surgery for metastatic spine disease	Scored out of a maximum of 10 points as follows: -Anemia: 1 point -Chronic lung disease: 1 point -Coagulopathy: 1 point -Electrolyte abnormalities: 1 point -Pulmonary circulation disorders: 2 points -Renal failure: 1 point -Malnutrition: 1 point -Emergent/Urgent case: 1 point -Anterior or combined surgical approach: 1 point	Stratified as quartiles as follows: -“No frailty”: MSTFI score of 0 -“Mild frailty”: MSTFI score of 1 -“Moderate frailty”: MSTFI score of 2 -“Severe frailty”: MSTFI score of 3 or higher
Modified 5-Item Frailty Index (mFI-5)	ACS NSQIP database with variables derived from Modified 11-Item Frailty Index (mF-11) which were originally derived from the 70-item Canada Study of Health and Aging Frailty Index (CSHA-FI)	Postoperative complications, mortality and unplanned 30-day readmission across all surgical specialties	Scored out of a maximum of 5 points as follows: -Hypertension requiring medication: 1 point -Partially-dependent or dependent functional status: 1 point -Insulin-dependent diabetes: 1 point -Chronic obstructive pulmonary disease: 1 point -Congestive heart failure: 1 point	Stratified as quartiles as follows: -"Robust": mFI-5 score of 0 -"Normal": mFI-5 score of 1 -"Frail": mFI-5 score of 2 -"Very frail": mFI-5 score of 3 or higher
mFI-11	ACS NSQIP databases with variables derived from the 70-item CSHA-FI	30-day postoperative complications and mortality across all surgical specialties	Scored out of a maximum of 11 points as follows: -History of myocardial infarction: 1 point -History of angina or percutaneous coronary intervention: 1 point -History of congestive heart failure: 1 point -History of transient ischemic attack or cerebrovascular accident without neurological deficit: 1 point -History of cerebrovascular accident with neurological deficit: 1 point -History of peripharal vascular disease or ischemic rest pain: 1 point -History of chronic obstructive pulmonary disease or pneumonia: 1 point -History of diabetes mellitus: 1 point -Nonindependent functional status: 1 point -History of impaired sensorium: 1 point -History of hypertension requiring medication: 1 point	Stratified as quartiles as follows: -"Robust": mFI-11 score of 0 -"Mild frailty": mFI-11 score of 1 -"Moderate frailty": mFI-11 score of 2 -"Severe frailty": mFI-11 score of 3 or higher
Hospital Frailty Risk Score (HFRS)	Hospital Episode Statistics inpatient database with variables identified by diagnostic codes from ICD, Tenth Revision	30-day mortality, length of hospital stay, and 30-day emergency readmission in elderly patients admitted in emergency settings	-Comprises of 109 variables that cover protein-energy malnutrition, dementia, and problems related to care-provider dependency -Scored via an algorithm having a complex weighting system derived from a logistic regression model with each of 109 variables awarded weight depending upon their strength of frailty prediction	Stratified as tertiles as follows: -"Low frailty score": HFRS < 5 -"Intermediate frailty score": 5 ≤ HFRS ≤ 14.9 -"High frailty score": HFRS ≥ 15
Revised-Risk Analysis Index (RAI-rev)	Veterans Affairs Surgical Quality Improvement Program (VASQIP) and ACS NSQIP databases with variables derived from the Minimum Data Set Mortality Risk Index-Revised instrument	30-day, 180-day, and 365-day mortality after VASQIP-elgible surgical procedures	–Scored out of a maximum of 81 points as follows: -Male sex: 3 points. -Age: Age is treated as a continuous variable, with points assigned based on age ranges. The points increase with advancing age, with additional points added for those with cancer. -Unintentional weight loss: 4 points -Poor Appetite: 4 points -Renal Failure: 8 points -Congestive Heart Failure: 5 points -Shortness of Breath: 3 points -Residence Other than Independent Living: 1 point -Functional Status and Cognitive Decline: This is a combined variable with points varying based on the level of functional dependence and the presence of cognitive decline. Higher points are assigned for greater dependence and cognitive faculty maintenance. –The RAI-rev uses 2 versions: RAI-A (retrospective: calculated after surgery using data from administrative databases) and RAI-C (prospective: designed to be administered before surgery, using a survey instrument). The scoring is identical except for the assessment of Activities of Daily Living (ADLs). The RAI-C allows for more granularity in measuring ADLs using a 5-point Likert scale for 4 domains, while the RAI-A relies on broader categories of functional status from administrative data	Stratified as quartiles as follows: -"Robust": RAI-rev ≤ 20 -"Normal": 21 ≤ RAI-rev ≤ 30 -"Frail": 31 ≤ RAI-rev ≤ 40 -"Very frail": RAI-rev ≥ 41
Johns Hopkins Adjusted Clinical Groups – Frailty Index (JHACG-FI)	Medicare claims dataset with variables identified by diagnostic codes from ICD-9	Health risk and healthcare costs for Medicare + Choice plan payment adjustments	–Scored out of a maximum of 5 points as follows: -Unintentional weight loss: 1 point -Exhaustion: 1 point -Low energy expenditure: 1 point -Low grip strength: 1 point -Slowed walking speed: 1 point	Stratified as tertiles as follows: -"Robust": JHACG-FI score of 0 -"Pre-frailty": JHACG-FI score of 1 or 2 -"Frailty": JHACG-FI score of 3 or higher

MSTFI, Metastatic Spinal Tumor Frailty Index; ACS NSQIP, American College of Surgeons – National Surgical Quality Improvement Program; ICD, International Classification of Diseases; mFI-5, Modified 5-Item Frailty Index; mFI-11, Modified 11-Item Frailty Index; CSHA-FI, Canada Study of Health and Aging Frailty Index; HFRS, Hospital Frailty Risk Score; RAI-rev, Revised-Risk Analysis Index; VASQIP, Veterans Affairs Surgical Quality Improvement Program; ADLs, Activities of Daily Living; JHACG-FI, Johns Hopkins Adjusted Clinical Groups – Frailty Index; ICD-9, ICD, Ninth Revision.

On the other hand, a modified 11-item frailty index (mFI-11) has been used since 2013 after being developed using the ACS-NSQIP database [29]. This frailty index was used initially to predict predict 30-day postoperative complications and mortality stratified for each surgical speciality including neurosurgery and orthopedic surgery [29]. The original investigation reported a gradual increase in postoperative complications and mortality across all surgical specialties for each unit increase in mFI-11 value [29]. It takes into account eleven variables and has been used in many different settings including orthopedic endovascular reconstruction, traumatic spinal cord injury (SCI), and pelvic fractures [[30], [31], [32]].

However, due to irregularities in recording of these variables by the ACS-NSQIP database, a more concise modification—mFI-5—was developed in 2018 that takes into account those variables which are more consistently administered in the ACS-NSQIP database [33]. It was used initially to predict postoperative complications, mortality and unplanned 30-day readmission after all surgical specialties including orthopedic surgical interventions [33]. An AUC of 0.72 for mortality, 0.62 for postoperative complications, and 0.59 for unplanned 30-day readmission were reported [33]. It has also been adopted across different settings including extradural spine tumors, cholelithiasis, and gynecologic surgery [[34], [35], [36]].

The Hospital Frailty Risk Score (HFRS) was developed and validated by Gilbert T et al. in 2018 using the Hospital Episode Statistics (HES) inpatient database [37]. Using 109 variables identified by ICD-10 codes, it was used initially to predict 30-day mortality, long length of hospital stay, and emergency readmission within 30-days in elderly patients admitted in emergency settings [37]. An AUC of 0.6 for 30-day mortality, 0.68 for a long length of hospital stay, and 0.56 for 30-day emergency readmission were reported [37]. Chronic obstructive pulmonary disease, heart failure, vascular surgery, and spine surgery are examples of clinical setting HFRS has been explored in [[38], [39], [40], [41]].

The revised Risk Analysis Index (RAI-rev) was developed using the VASQIP database and externally validated over the ACS-NSQIP database in 2020 [42]. It was initially developed to predict 30-day, 180-day, and 365-day mortality after VASQIP-eligible surgical procedures identified by Current Procedural Terminology (CPT) codes [42]. Using 7 variables, internal validation reported an AUC of 0.86 for 30-day mortality, 0.84 for 180-day mortality, and 0.81 for 365-day mortality [42]. External validation yielded an AUC of 0.87 for 30-day mortality [42]. RAI-rev has been explored across varying surgical settings including traumatic SCI, limbic flap reconstruction, and abdominal surgery [[43], [44], [45]].

Furthermore, the first version of Johns Hopkins Adjusted Clinical Groups (JHACG) was developed in 1992 using ICD-9 codes for health administrative data [46,47]. It has since been improved throughout the years and has acquired a binary classification model (JHACG-FI) that takes into account 10 variables [48]. Moreover, it has been adopted across diverse setting including cerebrovascular aneurysms, brain tumors, liver transplantation, and esophageal cancer [[49], [50], [51], [52]].

Components of frailty indices

Nononcologic comorbidities were taken into account by all 6 explored frailty indices, whereas 83% (n=5 of 6) frailty indices adopted performance function parameters (Table 3). None of the indices incorporated energy and exhaustion parameters. HFRS incorporated the highest number of frailty dimensions (6 out of 8) among the evaluated indices, covering 75% of our defined frailty spectrum (Table 3).

Table 3.

Variables adopted across frailty indices explored for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

Performance of frailty indices

The frailty indices were found to inconsistently predict outcomes of patients with MSD managed with surgery and / or radiotherapy (Table 4 & Supplementary Table 4). MSTFI and mFI-5 have been reported to provide a variable confidence of prediction potential for MSD surgery outcomes, with some studies reporting a significant association, [18,22] whereas others indicated an insignificant prediction potential [23,24]. Furthermore, mFI-11 could not predict postoperative complications and mortality, [26,27] whereas HFRS significantly predicted postoperative complications [16]. Moreover, RAI-rev and JHACG-FI were reported to independently predict postoperative complications and mortality after surgical intervention for MSD [3,21].

Table 4.

The performance of frailty indices for the prediction of outcomes in patients with metastatic spine disease managed with surgical intervention and / or radiotherapy.

Frailty indices	Performance of frailty indices
Metastatic Spinal Tumor Frailty Index (MSTFI)	It has been adopted in 6 studies [18,20,22,24,26,28]. 66% studies (n=4 out of 6) reported it to be significantly associated with postoperative outcomes (including postoperative complications and mortality) [18,20,22,28]. Furthermore, 17% studies (n=1 out of 6) reported it to be significantly associated with 3-months mortality, but not with 1-month mortality and postoperative complications [26], whereas 17% studies (n=1 out of 6) found it not to be significantly associated with both: postoperative complications and mortality [24].
Modified 5-Item Frailty Index (mFI-5)	It has been adopted in 5 studies [3,15,17,22,25]. 40% studies (n=2 out of 5) reported inability of mFI-5 to independently predict postoperative complications [3,15]. Moreover, mFI-5 poorly prognosticated 30-day mortality after surgery for MSD [17]. However, 20% studies (n=1 out of 5) reported mFI-5 to independently prgnosticate undefined postoperative complications [22]. Furthermore, it was not significantly associated with overall survival in patients with MSD managed with stereotactic body radiation therapy [25].
Modified 11-Item Frailty Index (mFI-11)	It has been adopted in 3 studies [23,26,27]. It was reported not to significantly predict postoperative outcomes (postoperative complications, mortality, and survival) in 67% (n=2 of 3) studies [26,27], whereas 33% (n=1 of 3) studies reported mFI-11 quartiles not to significantly predict overall survival in patients with MSD managed with surgical intervention and adjuvant radiotherapy [23].
Hospital Frailty Risk Score (HFRS)	It has been adopted in 2 studies [16,19]. Both studies reported it to be significantly associated with postoperative complications [16,19]. However, its association with postoperative mortality has not been explored.
Revised-Risk Analysis Index (RAI-rev)	It has been adopted in 1 study only, which reported it to independently predict postoperative complications and mortality [3].
Johns Hopkins Adjusted Clinical Groups – Frailty Index (JHACG-FI)	It has been adopted in 1 study only, which reported it to independently predict postoperative complications and mortality, albeit not across all patient groups stratified in terms of spinal segments [21].

In addition to this, Zakaria HM et al. reported mFI-5 quartiles not to be significantly associated with overall survival in MSD patients managed with SBRT as the primary treatment modality [25]. Moreover, Massaad E et al. found mFI-11 quartiles not to be significantly associated with overall survival in MSD patients managed with adjuvant radiotherapy [23].

Discussion

This systematic review identified 6 frailty indices explored in the context of MSD management: the MSTFI, mFI-5, mFI-11, HFRS, RAI-rev, and JHACG-FI. These indices varied in their components, with comorbidities being the most consistently included parameter across all indices. Notably, 83% of the indices incorporated performance function parameters, while none included energy and exhaustion measures. The HFRS emerged as the most comprehensive, covering 75% of the frailty spectrum as defined in this review. The performance of these indices in predicting outcomes for patients with MSD was inconsistent across studies, with some showing significant associations with postinterventional complications and mortality, while others failing to demonstrate predictive value.

The frailty indices identified in this review largely represented nonspecific tools developed for general surgical or geriatric populations, with most having been established several years ago. Only a single frailty index (MSTFI) was specifically designed for patients with MSD, highlighting a substantial gap in tailored assessment tools for this unique patient population. The remaining indices, including mFI-5, mFI-11, HFRS, RAI-rev, and JHACG-FI, were originally developed for broader patient groups and have been subsequently applied to spinal metastasis cases. This lack of specificity may partly explain the inconsistent performance of these indices in predicting outcomes for patients with MSD.

Components of frailty indices

The variability in components and performance of frailty indices reflected the complex nature of frailty assessment in patients with MSD. The consistent inclusion of comorbidities across all indices underscored their perceived importance in frailty assessment for this population. This aligns with the understanding that cancer patients often present with multiple comorbidities that can significantly impact surgical outcomes [7]. An understanding of the significance of comorbidities as covariates influencing outcomes of surgery in patients with MSD is also observed to be recognized amongst spine surgeons as a recent survey investigation revealed that spine surgeons had a consensus on high-risk cardiopulmonary diseases, renal failure, liver failure, and malnutrition portending a significant impact on the outcomes [9].

From the radiation oncology perspective, varying prognostic factors often take precedence, including epidural spinal cord compression (Bilsky grade), overall performance status, presence of pathological fractures, and the balance between oncologic and nononcologic disease burden [[53], [54], [55]]. None of the examined frailty indices incorporated these factors. This disparity in prognostic emphasis between surgical and radiation oncology perspectives highlights the need for more comprehensive, multidisciplinary assessment tools that can bridge these different clinical priorities.

A striking limitation of contemporary frailty assessment tools for patients with MSD was the exclusive focus on nononcologic comorbidities, neglecting cancer-specific factors such as metastatic burden (such as presence of visceral or brain metastases) and/or primary tumor characteristics. This omission may explain the limited predictive performance of these indices for patients with MSD, as metastatic burden and tumor characteristics are often primary determinants of treatment success and survival [56,57]. Furthermore, the relationship between frailty and spinal stability, as measured by the spinal instability neoplastic score (SINS), remains largely unexplored, representing a significant knowledge gap in treatment decision-making [[58], [59], [60]].

While most indices included performance function parameters, they failed to distinguish between oncologic and nononcologic contributions to performance status [61]. The absence of energy and exhaustion parameters in all reviewed indices is noteworthy, given the prevalence and impact of fatigue in cancer patients [62,63]. Although the HFRS offered a more comprehensive assessment, its complexity may limit practical application in clinical settings requiring rapid decision-making [64].

Future development of frailty indices should, therefore, aim to incorporate both surgical and radiation oncology perspectives, potentially weighting different components based on the intended treatment modality [65,66]. This could lead to more nuanced decision-making tools that not only predict complications but also guide treatment selection between surgical and radiotherapeutic approaches. It is also important to note that comprehensiveness in capturing frailty dimensions does not automatically confer superior predictive ability. Determining which index performs best would require direct head-to-head comparisons using the same patient cohort or a formal meta-analysis that could not be undertaken due to the scarcity of respective data.

The clinimetric limitations of existing frailty indices in spinal oncology were highlighted in a comprehensive review by MacLean MA et al., which found that most tools either lack or have uncertain content and construct validity when applied to this population [5]. Their analysis revealed that while most measures were objective and feasible, they demonstrated significant floor or ceiling effects that may limit their discriminative ability. This aligns with our findings regarding the inconsistent performance of these indices across studies. MacLean MA et al. also evaluated sarcopenia measures, which were not captured by the frailty indices identified in our review, suggesting another potential dimension of vulnerability assessment that remains unexplored in comprehensive frailty evaluation for patients with MSD.

Lastly, the lack of incorporation of social vulnerability data is an important limiting factor, with only 2 of the examined indices incorporating it [37,48]. There is growing evidence supporting socioeconomic disparities in patients with spine oncology, [67,68] and failure to incorporate these parameters may limit our understanding of the impact of frailty on outcomes.

Performance of frailty indices

Our review revealed inconsistent performance of frailty indices to predict outcomes of oncologic surgery for spinal metastases. This is in contrast to findings in other surgical fields as well as for degenerative spine ailment surgery where frailty indices consistently predicted adverse outcomes [[69], [70], [71], [72], [73]]. Our review also yielded inconsistent performance for outcomes prediction after radiotherapeutic management. This is also in contrast to findings observed for lung cancer, [74] prostate cancer, [75] and head and neck cancer, [76] where a statistically significant influence on radiotherapeutic outcomes was reported.

This discrepancy may be attributed to the unique challenges posed by this patient population. Unlike other oncological populations, patients with spinal metastases have advanced cancer with systemic effects, [77] which may confound traditional frailty assessments. Additionally, the complex interplay between tumor-related factors, treatment-induced complications, and baseline functional status in this population may potentially impact the predictive accuracy of indices. This necessitates a more specialized approach to frailty assessment for patients with MSD that goes beyond conventional geriatric indices.

Furthermore, this systematic review revealed scarcity of studies exploring frailty indices for prediction of outcomes for patients with MSD managed with radiation therapy. Radiotherapy-specific analysis was limited to only 2 studies (13%) [23,25]. While Massaad E et al. included patients receiving adjuvant radiotherapy, [23] the study did not segregate outcomes between surgery alone versus surgery with adjuvant radiotherapy, limiting conclusions about radiotherapy-specific effects. Zakaria HM et al. [25] focused exclusively on SBRT, a specialized and relatively less common palliative spine radiotherapy modality, finding no significant association between mFI-5 quartiles and overall survival.

The absence of studies examining conventional EBRT, which represents the more common approach to MSD, [78] is a substantial lacuna in the current literature. Moreover, the interpretation of radiotherapy outcomes in relation to frailty indices is complicated by inherent selection bias, as treatment decisions regarding radiation modality, dose, and fractionation are often influenced by the patient's underlying frailty status and life expectancy [79]. This creates a circular relationship where frailty influences treatment selection, which in turn affects outcomes, making it challenging to isolate the true predictive value of frailty indices directly for the outcomes in this context.

Interpretation and future directions

The inconsistent performance of contemporary frailty indices suggests caution in their application for clinical decision-making in MSD management. While potentially valuable, these tools should be part of a comprehensive assessment rather than used in isolation. Surgeons and multidisciplinary teams should adopt a more holistic approach incorporating patient-specific factors such as acuity, ambulatory status, and neurological function [80]. The MSTFI, though specifically developed for patients with MSD managed surgically, [28] has shown inconsistent performance across studies and should be used judiciously [24]. When using frailty indices for patient counseling, their limitations should be clearly communicated to ensure informed decision-making [81].

Our findings underscore the need for a more comprehensive, multidimensional frailty index tailored to patients with MSD. Future research should explore developing an index incorporating not only traditional frailty components but also MSD-specific variables such as neurological status, primary tumor characteristics, targetable mutations, prior oncological treatments, metastatic burden, and spinal instability. Such an approach could enhance predictive accuracy and potentially guide treatment decisions between surgical intervention and radiotherapy by weighing factors relevant to each modality alongside frailty measures. Prospective studies comparing this specialized index against existing measures would validate its clinical utility and explore integration into multidisciplinary tumor board discussions.

Limitations

This study has several limitations. The retrospective design of all included studies introduced potential bias and limited generalizability. Heterogeneity in outcome measures and statistical analyses complicated direct comparisons, while potential publication bias may have skewed our assessment. The limited representation of radiotherapy studies, particularly conventional EBRT, restricted generalizability to radiation oncology practice. Furthermore, there remains no consensus on what outcomes frailty indices should predict in cancer treatment. Although MacLean MA et al. reported that spine surgeons considered major complications, neurological recovery, and changes in performance status as most clinically relevant, [9] but universal agreement on predictive targets for frailty indices remains elusive—an area requiring future research.

Conclusion

This systematic review highlighted the variability and inconsistencies in frailty indices used for predicting outcomes in patients with MSD managed with surgery and / or radiotherapy. While several indices have been explored incorporating components such as comorbidities and functional status, their performance in predicting postinterventional complications and mortality has been inconsistent across studies. These findings underscored the unique challenges in assessing frailty in this patient population and the potential need for tailored, disease-specific frailty assessment tools.

Data availability statement

The data supporting the findings of this review are available within the article and its supplementary materials.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the author(s) used the Trinka AI web application for copy-editing and sentence structure optimization. After using this tool/service, the authors reviewed the content and take full responsibility for the content of the publication.

Author contributions

Rafael De la Garza Ramos and Ali Haider Bangash conceptualized and designed the study. Material preparation, data collection, and analysis were performed by Ali Haider Bangash, Rose Fluss, and Sertac Kirnaz. The first draft of the manuscript was written by Ali Haider Bangash, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.