Characterizing Health-Related Quality of Life by ambulatory status in patients with spinal metastases

Abstract

Study Design:

Retrospective review of prospective longitudinal data.

Objective:

To determine health-related quality of life (HRQL) utilities associated with specific ambulatory states in patients with spinal metastases: independent, ambulatory with assistance and non-ambulatory.

Summary of Background Data:

It is assumed that HRQL is aligned with ambulatory ability in patients with spinal metastases. Few studies have effectively considered these parameters while also accounting for clinical confounders.

Methods:

We used prospective longitudinal data from patients treated at one of three tertiary medical centers (2017–2019). HRQL was characterized using the Euroquol-5-dimension (EQ5D) inventory. We performed standardized estimations of HRQL stratified by ambulatory state using generalized linear modeling that accounted for patient age at presentation, biologic sex, follow-up duration, operative or non-operative management and repeated measures within the same participant.

Results:

We evaluated 675 completed EQ5D assessments, with 430 for independent ambulators, 205 for ambulators with assistance and 40 for non-ambulators. The average age of the cohort was 61.5. The most common primary cancer was lung (20%), followed by breast (18%). Forty-one percent of assessments were performed for participants treated surgically. Mortality occurred in 51% of the cohort. The standardized EQ5D utility for patients with spinal metastases and independent ambulatory function was 0.76 (95% CI 0.74, 0.78). Among those ambulatory with assistance, the standardized EQ5D utility was 0.59 (95% CI 0.57, 0.61). For non-ambulators, the standardized EQ5D utility was 0.14 (95% CI 0.09, 0.19).

Conclusions:

Patients with spinal metastases and independent ambulatory function have a HRQL similar to patients with primary cancers and no spinal involvement. Loss of ambulatory ability leads to a 22% decrease in HRQL for ambulation with assistance and an 82% reduction among non-ambulators. Given prior studies demonstrate superior maintenance of ambulatory function with surgery for spinal metastases, our results support surgical consideration to the extent that it is clinically warranted.

Level of Evidence:

3

Introduction

Spinal metastatic disease is a challenging condition to treat, given the limited life expectancy and potential for high morbidity and mortality documented in patients receiving operative and non-operative interventions for this condition.^1–13 Recent studies have shown that 1-year mortality in the setting of spinal metastases may be as high as 50%, even with careful selection of patients eligible for surgery.^4,9,10 In this light, the multidisciplinary teams responsible for developing plans of care for patients with spinal metastases must account for the risks and benefits of treatment strategies, whether operative or non-operative in nature, and effectively merge these with patient goals, expectations, and baseline function.^4,5

Patients with spinal metastases frequently express a desire to preserve independent function and maintain the highest quality of life possible, recognizing the limited life expectancy associated with the diagnosis.⁵ It is generally assumed that health-related quality of life (HRQL) is aligned with functional independence, ambulatory ability and daily pain in this population.^1,11,12,13 Nonetheless, while several studies document improvement in patient-reported HRQL outcomes after various treatments for spinal metastatic disease, the extent to which these findings are tied to ambulatory ability, as opposed to other parameters, is not well defined.^1,11,12,13 Furthermore, most of these investigations are compromised by selection, indication and surveillance bias—with estimates for outcomes at time-points of 1-year or more only applicable to the most robust patients with favorable disease characteristics.^1,11 While HRQL scores are recognized as being intimately tied to ambulatory ability, few studies have effectively considered these parameters while also accounting for the clinical and treatment-based characteristics that may confound results.

In this context, we used data from the Prospective Observational Study of Treatment for Spinal Metastases (POST)⁴ to determine HRQL utilities associated with specific, and mutually exclusive, ambulatory states encountered in patients with spinal metastases: independent ambulatory function, dependent ambulatory function (e.g. ambulatory with assistance; use of a cane or walker) and non-ambulatory function (e.g. wheelchair or bedbound). Based on prior research and our own clinical experience^{1,3,4,6,11,12,13}, we hypothesized that patients with independent ambulatory function would have superior HRQL scores when compared to dependent and non-ambulatory patients after adjusting for differences in clinical characteristics and operative, or non-operative, management.

Methods

We used data prospectively collected from 202 patients participating in the POST investigation at Brigham and Women’s Hospital, Massachusetts General Hospital and Dana Farber Cancer Institute who were enrolled from 2017–2019. The protocol for the POST investigation has previously been published.⁴ In brief, eligible participants included adult patients, 18 and older, presenting for treatment for spinal metastatic disease without prior intervention in the location of their metastatic involvement. Eligible patients were enrolled, with baseline information obtained within 72 hours of treatment initiation. Data collected included sociodemographic and clinical characteristics, primary tumor characteristics, serum albumin, ambulatory status at presentation, health-related quality of life characterized using the Euroquol-5-dimension (EQ5D) inventory^{1,11,12,13,14}, treatment strategy (e.g. operative vs. non-operative), and prognostic spinal metastases score including the Tokuhashi, Tomita, New England Spinal Metastasis Score (NESMS) and Spinal Instability and Neoplastic Score (SINS) criteria.⁴

Participants were then prospectively followed by study staff, with assessments at 1 month, 3 months, 6 months and 12 months following treatment initiation. If a participant was unable to fully participate in the surveys at a follow-up time-point due to incapacity, we used a healthcare proxy named at the time of enrollment to facilitate completion.⁴ Data obtained at each follow-up time-point consisted of surveillance for complications and readmissions since previous study-specific time-point, current ambulatory capacity and completion of the EQ5D. Participants were censored if they died during the course of the study. The data used for this investigation was finalized on August 1, 2020.

At all time-points in the study, ambulatory status was characterized as independently ambulatory, ambulatory with assistance (e.g. requiring use of walker, cane or crutches for majority of daily mobility), or non-ambulatory (e.g. wheelchair or bedbound due to pain or neurologic dysfunction, paralyzed, or only able to mobilize for transfers). Among the 202 participants in POST, we had available to us 675 completed EQ5D assessments, with 430 completed for independent ambulators, 205 for ambulators with assistance and 40 for non-ambulators, obtained at various stages in the disease course. EQ5D scores were transformed into utilities using normative values for the United States population.¹⁴ The EQ5D typically consists of a rating from 0.0 to 1.0, with 1.0 indicating perfect health and 0.0 reflecting death. However, participants can generate scores with a negative function, indicating a state they perceive to be worse than death.

We performed standardized estimations of HRQL stratified by ambulatory state using generalized linear modeling that accounted for patient age at presentation, biologic sex, follow-up duration, operative or non-operative management and repeated measures within the same participant.¹⁵ In these analyses, independent ambulatory function was used as the referent. We then developed 95% confidence intervals (CI) for each standardized estimation, taking into the account the number of measures available for each ambulatory state. Statistical significance was defined, a-priori, for measures with p-value <0.05 and mutually exclusive 95% CIs. All statistical tests were conducted with STATA v15.1 (STATA Corp., College Station, TX). Institutional investigational board approval was obtained prior to the commencement of this study. The POST investigation was funded by the Orthopaedic Research and Education Foundation (OREF).

Results

There were 202 patients enrolled in the POST investigation. The average age of this cohort was 61.5 (standard deviation [SD] 12.0), with 55% of the population identifying as male and 86% white (Table 1). The average number of co-morbidities per patient was 2.4, while the mean (SD) of the NESMS for the cohort was 1.7 (0.98). The mean (SD) Tokuhashi score was 8.4 (2.9), while it was 5.8 (2.6) for the Tomita score and 10.2 (3.1) for the SINS. The most common primary cancer was lung (20%), followed by breast (18%), prostate (14%) and renal (10%). The majority of patients (71%) had a serum albumin ≥3.5g/dL at presentation, with 58% independently ambulatory.

Table 1.

Characteristics of participants in the Prospective Observational Study for Spinal Metastasis Treatment (POST).

Characteristic	Value^
Age (mean, SD)	61.5 (12.0)
Male Sex	112(55)
Female Sex	90 (45)
White	174 (86)
Body Mass Index (mean, SD)	27.3 (6.2)
Number of Co-morbidities (mean, SD)	2.4 (0.9)
New England Spinal Metastasis Score (mean/SD)	1.7 (0.98)
Tokuhashi Score (mean/SD)	8.4 (2.9)
Tomita Score (mean/SD)	5.8 (2.6)
Spinal Instability Neoplastic (SINS) Score (mean/SD)	10.2 (3.1)
Primary Cancer &
Breast	35 (18)
Colon	3 (2)
Renal	20 (10)
Lung	39 (20)
Prostate	28 (14)
Treatment
Operative	80 (40)
Non-operative	122(60)
Bone Metastases	109 (54)
Visceral Metastases	106 (52)
Type of Lesion
Lytic	114 (57)
Blastic	41 (21)
Mixed (lytic/blastic)	45 (23)

^{^-}All values are presented as raw number and percentage except where otherwise noted.

^{& -}list is not exhaustive

Sixty-percent of the cohort received a non-surgical intervention, while 40% were treated operatively. Within the cohort treated operatively, 73% of procedures were performed in the thoracic spine. Eighty-one percent of operative events included a posterior fusion. Fifty-one percent of participants died within 1-year of presentation. No patients were lost to follow-up.

Among the 202 participants, there were 675 episodes where the EQ5D was measured. Forty-one percent (n=274) of assessments were performed for participants who had received a surgical intervention. For the cohort as a whole, the average EQ5D utility was 0.67 (SD 0.23). Within the cohort receiving surgery, the average utility at baseline was 0.51 (SD 0.28). Among the non-operative cohort, the average utility at presentation was 0.64 (SD 0.24). For independent ambulators, the average utility was 0.76 (SD 0.17), while it was 0.59 (SD 0.18) and 0.14 (SD 0.15) for patients who were ambulatory with assistance and non-ambulatory, respectively.

Following generalized linear modeling that accounted for potential confounders, a status of ambulatory with assistance was associated with a 0.16 reduction (−0.16; 95% CI −0.18, −0.13; p<0.001) in EQ5D utility. In the same analysis, non-ambulatory status was associated with a 0.58 (−0.58; 95% CI −0.63, −0.53; p<0.001) reduction in EQ5D utility. Other major drivers of changes in the EQ5D utility were operative intervention (−0.06; 95% CI −0.08, −0.03; p<0.001) and time from presentation (e.g. duration of follow-up; p<0.001 for all; Table 2).

Table 2:

Results of adjusted Generalized Linear Modeling for Health Related Quality of Life by Euroquol 5-dimension (EQ5D) utility based on ambulatory status and co-variates.

	Regression Estimate	95% Confidence Interval	p-value
Ambulatory Status
Independently ambulatory	Ref
Ambulatory with assistance	−0.16	−0.18, −0.13	<0.001
Non-ambulatory	−0.58	−0.63, −0.53	<0.001
Age	0.001	0.0, 0.002	0.06
Male sex	0.04	0.02, 0.07	0.001
Surgery	−0.06	−0.08, −0.03	<0.001
Duration of Follow-up
Baseline enrollment	Ref
1-month	0.07	0.04, 0.10	<0.001
3-months	0.09	0.05, 0.13	<0.001
6-months	0.10	0.06, 0.14	<0.001
12-months	0.13	0.08, 0.18	<0.001
Constant	0.63	0.56, 0.70	-

The standardized assessment for EQ5D utility for patients with spinal metastases and independent ambulatory function was 0.76 (95% CI 0.74, 0.78; Table 3). Among those ambulatory with assistance, the standardized EQ5D utility was 0.59 (95% CI 0.57, 0.61). For those who were non-ambulatory, the standardized EQ5D utility was 0.14 (95% CI 0.09, 0.19).

Table 3.

Point estimates and 95% confidence intervals (CI) for Euroquol 5-dimension (EQ5D) utilities stratified by ambulatory Status.

Ambulatory Status	Observations	Utility	Standard Deviation	95% CI
Independently ambulatory	430	0.76	0.17	0.74, 0.78
Ambulatory with assistance	205	0.59	0.18	0.57, 0.61
Non-ambulatory	40	0.14	0.15	0.09, 0.19

Discussion

Investigating the relationship between HRQL and clinical characteristics in patients with spinal metastases, particularly within the context of ambulatory capacity and functional independence, is challenging given that ambulatory status is often fluid across the disease state and declination in function may only transpire many months following treatment initiation. Previous studies that have addressed this issue to some degree have been impaired by small sample size, selection bias and inability to associate HRQL with ambulatory ability or other measures of functional independence.^1,11,12,13

In the current analysis, we have remedied many of these shortcomings by using prospectively-collected, longitudinal data from patients participating in the recent POST investigation (2017–2019), taking into account repeated estimates of ambulatory ability and HRQL as measured by the EQ5D. The 202 participants in POST were treated in one of three tertiary referral centers using contemporary operative and non-operative techniques and were followed for a period of 1-year after presentation, or until they passed away, with no individuals lost to follow-up. This analytic approach allowed us to control for the influence of selection and indication bias regarding treatment, duration of follow-up and patient longevity following treatment initiation, as well as repeated measures within the same individual. Furthermore, the average score and SD of the clinical utilities we considered, including the NESMS, Tokuhashi, Tomita and SINS, indicate that we captured a substantial range of the clinical variation typically encountered in patients presenting with spinal metastases. This may indicate that our findings are more representative of the typical patient with spinal metastatic disease and therefore have a broader capacity for generalization.

In the current work, we determined that patients with spinal metastases and independent ambulatory function have an EQ5D utility of 0.76 (95% CI 0.74, 0.78). Those who decline in function to the point where they require assistive devices, report a 22% reduction in HRQL (EQ5D utility=0.59; 95% CI 0.57, 0.61). Meanwhile, patients who become non-ambulatory experience an 82% reduction in HRQL (EQ5D utility=0.14; 95% CI 0.09, 0.19). For contextualization, other studies have reported that the EQ5D utility for a proximal humerus fracture is 0.88¹⁶, while that of adult cervical deformity and symptomatic spinal stenosis prior to surgery are 0.51¹⁷ and 0.40¹⁸, respectively (Table 4). Diseases of the circulatory system, depending on severity, have been reported to have EQ5D utilities in the range of 0.30–0.97.¹⁹ Among oncologic conditions, the EQ5D utility for head and neck cancer was estimated at 0.87²⁰, with similar findings for patients with esophageal cancer (0.80)²¹ and non-small cell lung cancer (0.76)²². These findings are particularly sobering for clinicians who treat this daunting patient population and highlight the paramount importance of prioritizing indicated interventions (whether operative or non-operative) that seek to preserve ambulatory status.

Table 4:

Our estimates of Euroquol 5-dimension (EQ5D) utility based on ambulatory status in the context of other studies examining health-related quality of life in other cancer-based, spine and musculoskeletal conditions.

Study	Condition	Sample Size	Utility
Spross et al16	Proximal humerus fracture	192	0.88
Davies et al20	Head and Neck cancer	2,065	0.87
Doherty et al21	Esophageal cancer	199	0.80
Current study	Spinal metastases: Independently ambulatory	430	0.76
Versteeg et al1	Spinal metastases: Non-operative cohort	84	0.66
Current study	Spinal metastases: Ambulatory with assistance	205	0.59
Versteeg et al1	Spinal metastases: operative cohort	136	0.52
Smith et al17	Adult cervical deformity	121	0.51
Miyazaki et al11	Spinal metastases: operative cohort at 12-months	-	0.45
De Ruiter et al12	Spinal metastases: Presentation	113	0.44
Burgstaller et al18	Lumbar spinal stenosis pre-treatment	364	0.40
Morgen et al13	Spinal metastases: operative cohort at enrollment	69	0.28
Current study	Spinal metastases: Non-ambulatory	40	0.14
Miyazaki et al11	Spinal metastases: operative cohort at enrollment	31	0.04

Many of the clinical characteristics in our study were comparable to those from existing literature regarding patients with spinal metastases, which further supports the external validity of these results. For example, the average age of the cohort under study, prevalence of primary tumors, and patient longevity, were similar those in other published studies.^{1,2,8,9,10,12,13} We posit that our finding of association between operative intervention and lower EQ5D utility should not be interpreted that surgery is associated with inferior HRQL outcomes, either within this cohort of patients, or in general. Rather, we believe this finding is instead reflective of selection bias, in that patients who received surgery had lower EQ5D scores at baseline than the non-operative cohort. Analogous findings are encountered in the study of Morgen et al., which noted that patients treated surgically had a baseline EQ5D utility of 0.28 compared to a value of 0.42 among those receiving non-operative care.¹³ Similarly, in the work of Versteeg et al., baseline EQ5D was 11 points lower (0.54 vs 0.65) in patients selected for operative rather than non-operative management.¹ In this context, our findings regarding increased EQ5D scores at each successive time-point in the study are not only reflective of the benefits of treatment, but also underscore a surveillance bias in that the most functional patients with superior post-treatment trajectories survived to be evaluated at later time-points in the study. To address this possibility, we adjusted for both the selection bias of surgical intervention and the surveillance bias associated with patient survival by including these factors as co-variates in our generalized linear model, thus accounting for multiple measures at different time-points within the same individual.

We believe that our results have an important message for patients with spinal metastases, as well as their families, treating clinicians and insurance carriers. Foremost, and most simply, our results provide a context and scale by which patients and providers can understand the impact of ambulatory status in spinal metastases and its relevance to other musculoskeletal, oncologic and other chronic disease states. Based on the estimates provided in Table 4, we believe it is imperative to recognize that with indicated treatment (operative or non-operative depending on the clinical context) that can preserve independent ambulatory function, patients with spinal metastases enjoy a HRQL (0.76) proximate to that of other cancer patients without spinal involvement (0.76–0.87). Furthermore, loss of independent ambulatory ability leads to dramatic declinations in quality of life with a 22% decrease in patients who require assistive devices and a more dramatic 82% reduction in those who become wheelchair or bedbound. Such information can help inform shared-decision making discussions as the impact of different treatment strategies on quality and quantity of life can now be taken into account. Several previous studies have demonstrated superior maintenance of ambulatory function with surgical intervention for spinal metastases.^6,7,10 We believe that the results of those studies paired with our current findings support strong consideration for surgical intervention in this clinical context to the extent that this is practicable. Finally, we believe that our standardized estimates of HRQL by ambulatory status are sufficiently generalizable and robust, given our sample size and clinical characteristics of the population under study, to serve as metrics in external cost-effectiveness and cost-utility analyses.

We recognize several limitations associated with this work. First, patients were treated in a single city in the Northeast United States and all three referral centers are teaching partners of a single medical school that share several overlapping training programs and educational initiatives. This may impart the potential for a clustering bias that cannot be controlled for. In addition, while over 200 patients were enrolled in the POST study, there may still exist a degree of restricted clinical variation, particularly when considering that a plurality of cases were lung cancer and the fact that over 60% of the spinal metastases encountered in our study derived from only one of four tumor types (lung, breast, renal, prostate). We do contend, however, that the wide variation reflected in the mean and SD of the scoring utilities (e.g. NESMS, SINS, Tomita, Tokuhashi) generated for the sample are reassuring in this regard. Nonetheless, our estimates should be replicated in other populations where the sociodemographic and clinical characteristics vary to a significant degree before complete investment in our determinations is possible.

In conclusion, we found that the standardized EQ5D utility for patients with independent ambulatory function in the setting of ambulatory metastases was 0.76. Those who can maintain independent ambulatory function have a HRQL similar to patients with primary cancers and no spinal involvement. However, loss of independent ambulatory ability leads to dramatic declinations in HRQL, with a 22% decrease in EQ5D utility for ambulation with assistance and a 82% reduction for those who become non-ambulatory. Given that prior studies have demonstrated superior maintenance of ambulatory function with surgical intervention for spinal metastases, our results support consideration for surgical intervention to the extent that it is clinically warranted.