Is Ambulatory Status a Prognostic Factor of Survival in Patients with Spinal Metastases? An Exploratory Meta-analysis

Abstract

This study was conducted to identify the influence of ambulatory status prior to treatment on survival of patients with spinal metastases. Two investigators independently retrieved relevant electronic literature in PubMed, Embase, and Cochrane Library databases, to identify eligible studies. Effect estimates for hazard risk (HR) were extracted and synthesized through fixed-effects or random-effects models as appropriate. A total of 17 eligible studies were identified, with an accumulated number of 3962 participants. HR from 14 studies regarding comparison between ambulatory versus non-ambulatory groups were pooled using a random-effects model, and statistical significance was presented for the pooled HR (HR = 1.96; 95% confidence interval [CI], 1.65-2.34). In subgroups of mixed primary tumor and lung cancer, ambulatory status was considered to be a significant prognostic factor (P < 0.05), while in the subgroup of prostate cancer it was not (HR = 1.72; 95% CI, 0.79-3.74). HR from 4 studies related to comparison between Frankel E versus Frankel C-D were pooled using a fixed-effects model, which revealed statistical significance (HR = 1.73; 95% CI, 1.27-2.36). Ambulatory status is a significant prognostic factor in patients with spinal metastases. However, in patients with primary prostate cancer, the prognostic effect of ambulatory status has not yet been confirmed to be significant.

Keywords: Ambulatory status; Overall survival; Prognostic factor; Spinal metastasis.

Figure 1

Flowchart of eligible literature selection.

Figure 2

Forest plots presenting combined effect estimates (hazard ratios [HR]) of survival in spinal metastatic patients: (A) ambulatory versus non‐ambulatory, a subgroup analysis based on the histology of primary tumor; and (B) Frankel E versus Frankel C&D.

Figure 3

Plots of sensitivity analysis for 14 studies included in the forest plot A (A) and studies in subgroups of primary prostate cancer (B) and primary lung cancer (C).

Figure 4

Funnel plot (A) and Egger's publication bias plot (B) presenting publication condition of 14 studies related to patients ambulatory versus non‐ambulatory before treatment, which were included in forest plot A; Funnel plot (C) and Egger's publication bias plot (D) presenting the risk of bias across 4 studies related to patients without neurological defect (Frankel grade E) versus with neurological defect (including Frankel grade C‐D) before treatment which were included in forest plot B.