The Spine Journal 17 (2017) 689–708

Clinical Study

Prognostic factors in patients with spinal metastasis: a systematic review

and meta-analysis
Panya Luksanapruksa, MDa, Jacob M. Buchowski, MD, MSb,*, William Hotchkiss, MDb,
Sasima Tongsai, PhDc, Sirichai Wilartratsami, MDa, Areesak Chotivichit, MDa
a
Department of Orthopedic Surgery, Faculty of Medicine Siriraj Hospital, Mahidol University, 2 Wanglang Rd Bangkoknoi, Bangkok 10700, Thailand
b
Department of Orthopedic Surgery, Barnes-Jewish Institute of Health, Washington University in St. Louis, 425 S. Euclid Ave, Campus Box 8233, St. Louis,
MO 63110, USA
c
Office for Research and Development, Faculty of Medicine Siriraj Hospital, Mahidol University, 2 Wanglang Road Bangkoknoi, Bangkok 10700, Thailand
Received 5 May 2016; revised 10 November 2016; accepted 9 December 2016

Abstract BACKGROUND CONTEXT: Incidence of symptomatic spinal metastasis has increased owing
to improvement in treatment of the disease. One of the key factors that influences decision-making
is expected patient survival. To our knowledge, no systematic reviews or meta-analysis have been
conducted that review independent prognostic factors in spinal metastases.
PURPOSE: This study aimed to determine independent prognostic factors that affect outcome in
patients with metastatic spine disease.
STUDY DESIGN: This is a systematic literature review and meta-analysis of publications for prog-
nostic factors in spinal metastatic disease.
PATIENT SAMPLE: Pooled patient results from cohort and observational studies.
OUTCOME MEASUREMENT: Meta-analysis for poor prognostic factors as determined by hazard
ratio (HR) and 95% confidential interval (95% CI).
METHODS: We systematically searched relevant publications in PubMed and Embase. The
following search terms were used: (“‘spinal metastases’” OR “‘vertebral metastases’” OR ““spinal
metastasis” OR ‘vertebral metastases’) AND (‘“prognostic factors”’ OR “‘survival’”). Inclusion
criteria were prospective and retrospective cohort series that report HR and 95% CI of independent
prognostic factors from multivariate analysis. Two reviewers independently assessed all papers.
The quality of included papers was assessed by using Newcastle-Ottawa Scale for cohort studies
and publication bias was assessed by using funnel plot, Begg test, and Egger test. The prognostic
factors that were mentioned in at least three publications were pooled. Meta-analysis was
performed using HR and 95% CI as the primary outcomes of interest. Heterogeneity was assessed
using the I2 method.

FDA device/drug status: Not applicable.
Author disclosures: PL: Nothing to disclose. JMB: Royalties: Wolters
Kluwer Health, Inc. (A), Globus Medical, Inc. (E), K2M, Inc. (C); Con-
sulting: CoreLink, Inc. (B), Globus Medical, Inc. (B), K2M, Inc. (B),
Medtronic, Inc. (C), Stryker, Inc. (B); Speaking and/or Teaching Arrange-
ments: Broadwater/Vertical Health (Spine Surgery LIVE) (B), DePuy Synthes
(C), Globus Medical, Inc. (C), Orthofix (C), Stryker, Inc. (C); Fellowship
Support: OMeGA (D, Paid directly to institution/employer), AO North America
(F, Paid directly to institution/employer), OREF (D, Paid directly to institution/
employer), outside the submitted work. WH: Nothing to disclose. ST: Nothing
to disclose. SW: Nothing to disclose. AC: Nothing to disclose.
No funds were received in support of this work.
There are no relevant financial activities outside the submitted work.
The disclosure key can be found on the Table of Contents and at
www.TheSpineJournalOnline.com.
* Corresponding author. Department of Orthopedic Surgery, Barnes-
Jewish Institute of Health, Washington University in St. Louis, 425 S. Euclid
Ave, Campus Box 8233, St. Louis, MO 63110, USA. Tel.: +1 (314) 747 4950;
fax: +1 (314) 747 2599.
E-mail address: buchowskij@wustl.edu (J.M. Buchowski)

http://dx.doi.org/10.1016/j.spinee.2016.12.003
1529-9430/© 2016 Elsevier Inc. All rights reserved.
690 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

RESULTS: A total of 3,959 abstracts (1,382 from PubMed and 2,577 from Embase) were
identified through database search and 40 publications were identified through review of cited
publications. The reviewers selected a total of 51 studies for qualitative synthesis and 43 studies for
meta-analysis. Seventeen poor prognostic factors were identified. These included presence of a
neurologic deficit before surgery, non-ambulatory status before radiotherapy (RT), non-ambulatory
status before surgery, presence of bone metastases, presence of multiple bone metastases (>2 sites),
presence of multiple spinal metastases (>3 sites), development of motor deficit in <7 days before
initiating RT, development of motor deficit in <14 days before initiating RT, time interval from
cancer diagnosis to RT <15 months, Karnofsky Performance Score (KPS) 10–40, KPS 50–70,
KPS<70, Eastern Cooperative Oncology Group (ECOG) grade 3–4, male gender, presence of
visceral metastases, moderate growth tumor on Tomita score (TS) classification, and rapid growth
tumor on TS classification.
CONCLUSIONS: Seventeen independent poor prognostic factors were identified in this study. These
can be categorized into cancer-specific and nonspecific prognostic factors. A tumor-based prognos-
tic scoring system that combines all specific and general factors may enhance the accuracy of survival
prediction in patients with metastatic spine disease. © 2016 Elsevier Inc. All rights reserved.

Keywords: Systematic review; meta-analysis; survival; prognostic factors; spinal metastases; prognostic score

Introduction Materials and methods

The spine is one of the most common sites of metastatic This systematic review and meta-analysis was con-
disease. The prevalence of symptomatic spinal metastasis has ducted according to the Preferred Reporting Items for
increased owing to treatment improvements and longer patient Systematic Reviews and Meta-Analyses protocol [7].
survival. The current treatments for spinal metastasis are ra-
diotherapy (RT), systemic chemotherapy, and surgery. One
of the key factors that could assist in decision-making is ex- Data source and search strategy
pected survival of the patient.
We searched Embase and PubMed for all studies before
Four scoring systems provide survival estimates for pa-
May of 2015 involving survival factors for patients with spinal
tients with spinal metastasis. The revised Tokuhashi [1], Tomita
metastasis. The following search terms were used: (“‘spinal
[2], van der Linden [3], and Bauer [4] scoring systems use dif-
metastases’” OR “‘vertebral metastases’” OR ““spinal me-
ferent factors and weigh these factors to calculate expected
tastasis” OR ‘vertebral metastases’) AND (‘“prognostic
survival. The revised Tokuhashi scoring system uses six factors
factors”’ OR “‘survival’”). The search limits were English
on a 15-point scale: a general condition or performance score,
language, studies conducted in humans, and full text avail-
number of extraspinal bone metastases, number of vertebral
able. The inclusion criteria were prospective and retrospective
body metastases, presence or removability of metastases to major
cohort series that reported survival factors (hazard ratio [HR]
internal organs, primary cancer site, and Frankel neurologic
and 95% confidential interval [95% CI]) in adult patients with
condition. The Tomita scoring system uses three factors on a
spinal metastases. Exclusion criteria were no survival factor
10-point scale: growth rate of primary tumor, presence or treat-
report, mixed report with other bone metastases, duplicated
ability of visceral metastases, and the number of bone metastases.
report, no multivariate analysis, fewer than 30 patients, and
The van der Linden scoring system uses three factors on a 6-point
another type of publication including case report, review article,
scale: Karnofsky Performance Score (KPS), primary tumor, and
or technical report. Additional articles were identified from
presence or absence of visceral metastases. Using different
the reference lists of the retrieved studies that were relevant
scoring systems for the same patient can result in different ex-
to survival in spinal metastases. Both reviewers (PL, SW) in-
pected survival periods that lead to different treatment strategies.
dependently screened abstracts and titles after removing
In addition, additional other survival factors were not in-
duplicated publications. Then, full-paper readings were per-
cluded in the previous scoring systems, which could offer help
formed to determine final inclusion. Disagreements were
to decision-makers including postoperative ambulatory status
resolved by discussion for consensus.
[5], albumin level [6], and others.
To our knowledge, no systematic review or meta-analysis
that reviews survival factors for patients with spinal metasta-
Quality assessment
ses has been conducted. The purpose of this study is to review
independent survival factors in spinal metastases from cohort Quality assessment of included studies was performed
studies. independently by two reviewers (PL, SW) using the
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
Context
The authors performed a systematic review to determine
negative prognostic factors in patients with spinal
metastases.
Contribution
They found 17 independent negative prognostic factors.
Implications
While many of the factors are known and several are in-
corporated into existing classification systems (Tokuhashi,
Tomita, etc.), this study provides pooled information that
should be quite helpful for patient education/discussion of
prognosis. The tables specifically might prove quite
beneficial.
Newcastle-Ottawa Scale for cohort studies with total score
ranges of 0–9 calculated from three major categories includ-
ing selection, comparability, and outcome [8].
Data extraction and outcome assessment
This review focused on prognostic factors for overall sur-
vival in patients with spinal metastasis. Both reviewers
independently extracted data including first author, name of
journal, year of publication, data collection year, treatment,
number of patients, type of study (prospective or retrospec-
tive), mean or median age of the sample, type of tumor,
independent survival factors, HR, and 95% CI from multi-
variable analyses only. Disagreements were discussed until
consensus was achieved.
Data analysis and statistical analysis
Hazard ratio was used as the primary effect estimate in
the meta-analysis. However, only HRs and 95% CI of the
variables reported as significant predictors in at least three
papers were pooled in meta-analysis. Poor prognostic factors
were the factors that had negative effect to the survival of
patients with spinal metastasis. The outcome was defined as
the risk of a poor prognostic factor. Risk of favorable
prognostic factor was recalculated to the risk of poor
prognostic factor. In case the reference category for the
exposed category differed between studies, the HRs were
recalculated, specifying the lowest risk category as the
reference. Statistical heterogeneity was assessed using Co-
chrane Q test, with a p-value set at .1 for significance.
Heterogeneity between trials was evaluated based on
inconsistency (I2) index, and substantial heterogeneity
was represented by an I2 value greater than 50%. The
691
fixed-effects model was used when the effects were assumed
to be homogenous. In the presence of heterogeneity, we
used a random-effect model. Sensitivity analysis was per-
formed by omitting each study at a time to investigate its
effect on the overall meta-analysis result. We rejected the
studies that caused higher statistical heterogeneity (I2>90).
A p-value <.05 was considered statistically significant.
Publication bias was assessed using a funnel plot, Begg
test, and Egger test. A trim-and-fill method was used to
estimate pooled HR after adjusting for funnel plot asymme-
try arising from publication bias. The interobserver reliability
between two reviewers was assessed by kappa statistic.
Meta-analysis was performed using meta package (R Devel-
opment Core Team, 2015, Vienna, Austria) version 3.2.2.
Results
Included studies
A total of 3,959 abstracts were identified through data-
base searching (1,382 from PubMed and 2,577 from Embase),
and 40 additional publications were identified through review
of cited publications. There were 642 duplicate publica-
tions, and thus a total of 3,357 unique abstracts were screened.
Of these abstracts, 135 were selected for full paper review.
A total of 84 articles were excluded for the following reasons:
38 had no survival factor reported, 23 had no multivariate anal-
ysis, 2 had duplicate reports, 4 had results mixed with other
bone metastases, 2 had no significant factor after multivari-
ate analysis, 1 had fewer than 30 patients, six were other types
of publications (case report, review article, or technical report),
and 8 did not report HR or a 95% CI. The reviewers se-
lected a total of 51 studies for qualitative synthesis and 43
studies for meta-analysis. The flow diagram of the literature
search is shown in Fig. 1. The kappa statistics of article screen-
ing, data extraction, and quality assessment were 0.98 (95%
CI, 0.96–1.00), 1.00, and 0.90 (95% CI, 0.80–1.00)
respectively.
Characteristics and quality of the included studies
A total of 13,666 patients were enrolled from 51 in-
cluded studies. Most of the included studies were retrospective
(48 of 51) [5,6,9–53], conducted in a single institute (31 of
51), and published between 2005 and 2015. Three studies were
prospective cohort studies [3,54–56]. The number of en-
rolled patients ranged from 30 to 2,029 patients. The age of
enrolled patients ranged from 23.15 to 72.3 years, and median
survival ranged from 4.5 to 22 months. Forty-nine percent
of included patients (25 of 51) reported mixed primary site
of tumors. Additionally, the percentage of studies that re-
ported patients who underwent only RT, only surgery, or who
received unclear treatment were 45.09%, 43.13%, and 11.76%,
respectively. The average Newcastle-Ottawa Scale score was
6.98 (range 6–8). Characteristics of the included studies are
shown in Table 1.
692 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
Fig. 1. The Preferred Reporting Items for Systematic Reviews and Meta-
Analyses flow diagram of literature search.
Prognostic factors
There were poor prognostic factors specific to each type
of tumor. Breast cancer included negative estrogen receptor
[40,43], negative progesterone receptor [43], negative hor-
monal receptor [43], poorly or undifferentiated pathology [11],
and shorter course RT (1×8 Gy or 5×4 Gy) [50]. Hepatocel-
lular carcinoma included biologically effective dose Gy10<38
and uncontrolled primary hepatocellular carcinoma [13]. Lung
cancer included biologically effective dose <40 Gy, tissue other
than adenocarcinoma, and no use of epidermal growth factor
receptor-targeted treatment in adenocarcinoma [20]. Non–
small cell lung cancer included non-adenocarcinoma pathology
[51], and renal cell carcinoma included no zoledronic acid
treatment [45] and Fuhrman grade 4 [58]. Detail of specific
factors is shown in Table 2.
The nonspecific poor prognostic factors included age, sys-
temic disease, non-ambulatory status before RT, preoperative
non-ambulatory status, American Society of Anesthesiolo-
gists (ASA) status, extraspinal bone metastases
[19,20,23,29,30,32,33,36,37,39,44,50,51,53,55], number of
spinal metastases, abnormal blood test, comorbidities, pre-
vious chemotherapy, primary tumor site, interval from
diagnosis to spinal metastases [19,29,30,41], time develop-
ing motor deficits before RT, interval from diagnosis to RT,
preoperative neurologic deficit, severe pain, KPS, Eastern Co-
operative Oncology Group (ECOG) grade, no adjuvant therapy,
RT course, revised Tokuhashi score (RTS), Tomita score (TS),
gender, and visceral metastases. However, each group had their
own variables. The details of nonspecific factors are shown
in Table 3, and prognostic factors comparing each primary
tumor site are shown in Table 4.
There were nine poor prognostic factors reported from
single studies, including admission through emergency room
[11], increased time from diagnosis of cancer to surgery [11],
time from primary diagnosis to operation <6 months [55],
pelvis metastases compared with lower limb metastases
[55], local disease progression after stereotactic radiosur-
gery [41], progression of systemic disease at time of
stereotactic radiosurgery [41], preoperative RT [56], opera-
tive strategy other than en bloc resection [55], and postoperative
non-ambulatory status [27].
There were multiple independent prognostic factors in the
eight studies that were excluded because of no reported HR
or 95% CI. Demirci et al. reported on nasopharyngeal cancer
metastases including age, American Joint Committee on
Cancer T classification, and tumor response [59]. Drzymalski
et al. reported on prostate cancer metastases including pres-
ence of additional metastasis at the diagnosis of spinal
metastasis, longer duration between the diagnosis of pros-
tate cancer to spinal metastasis, and higher prostate-specific
antigen level at the diagnosis [60]. Ogihara et al. reported on
non–small cell lung cancer metastases including serum albumin
level below 3 g/dL, serum calcium more than 10.5, and ECOG
Performance Status 3–4 [61]. Rades et al. reported on unknown
primary cancer metastases including non-ambulatory status
before RT, other bone metastases at the time of RT, time to
develop motor deficit before RT more than 7 days, and vis-
ceral metastases at the time of RT [62]. There were four studies
that reported on mixed tumor metastases including KPS, vis-
ceral involvement, systemic treatment, rapid primary tumor
and visceral metastases, increasing age, and an RTS below
7 [3,52,63,64].
Meta-analysis result
As noted above, 17 poor prognostic factors were
identified. These included the presence of a neurologic
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 693

Table 1
Characteristics of the included studies
Mean
Publication Collection No. age Median survival Tumor
Author Nos. year year of pts. (y) (95% CI) (mo) type Treatment Type Prognostic factors HR 95% CI
Arrigo 7 2011 1999–2009 200 58.9 8 (6.9–10.3) Mixed Surgery R Non-ambulatory 2.355 1.517–3.658
et al. Charlson comorbidity 2.955 1.341–6.512
[9] index ≥2
Radio-resistant tumor 2.557 1.3672–9.912
Bollen 7 2013 2001–2010 106 59 10.7 (0.2–107.5) Mixed Surgery R KPS 10–40 2.7 1.1–6.6
et al. Tomita fast-growing 3.1 1.6–6.2
[10] tumors
Visceral metastases 1.7 1.0–2.9
Cahill 8 2011 1986–2005 379 72 15.7 (11.9–18.5) Breast Surgery R Admission via 1.53 1.20–1.97
et al. emergency room
[11] Time cancer diagnosis to 0.97 0.95–0.99
surgery (per 1 y)
Poorly or 1.49 1.14–1.95
undifferentiated
histology
Chen 8 2010 2001–2007 41 23.15 10.36 (8.22-2.49) HCC Mixed R Serum albumin>37 g/dL 0.295 0.106–0.819
et al. LDH>200 U/L 5.626 1.562–20.265
[12]
Choi 8 2014 1992–2012 192 56 4.5 (3.7–5.3) HCC RT R No extrahepatic 0.571 0.391–0.835
and metastases other than
Seong bone
[13] ECOG status 3–4 1.895 1.302–2.757
Biologically effective 0.536 0.393–0.751
dose Gy10>53
Uncontrolled primary 3.585 0.383–0.751
HCC
Chong 7 2012 2002–2010 105 58.3 6 (0–45) Mixed Surgery R No postoperative 3.23 1.80–5.77
et al. adjuvant therapy
[14] Number of spinal 1.94 1.10–3.43
metastases≥3 levels
Crnalic 8 2012 2003–2010 68 72.3 5 (0.3–59) Prostate Surgery R KPS 50–70 3.97 1.57–10.04
et al.
[15]
Dardic 6 2014 2005–2010 196 66 7 (0.16–70) Mixed Mixed R KPS 50–70 2.07 1.44–2.96
et al. KPS 10–40 2.7 1.70–4.30
[16] Multiple spinal 1.49 1.04–2.14
metastases
Tomita fast growth tumor 2.05 1.44–2.92
Treatable visceral 1.93 1.31–2.85
metastases
Untreatable visceral 4.77 2.98–7.63
metastases
Han 8 2015 2003–2012 30 52.2 17 (N/A) RCC Surgery R Revised Tokuhashi score 0.225 0.063–0.796
et al. (>10)
[17]
Hosono 6 2005 1985–2001 165 N/A N/A Mixed Surgery R Present paresis 1.79 1.11–2.96
et al. Present pain 2.53 1.12–7.26
[18] Poor primary tumor site 2.36 1.67–3.35
Kataoka 6 2012 1990–2008 143 61 22 (1–127) Mixed Mixed R Extraspinal bone 1.75 1.04–2.79
et al. metastases
[19] Disease free before 1.77 1.02–2.34
metastases<12 mo
Tomita: mod. growth 1.797 1.12–3.39
tumor
Tomita: rapid growth 6.802 3.62–11.5
tumor
Major organ metastases 2.01 1.09–2.34
Komatsu 8 2012 2004–2009 132 63 4.8 (N/A) Lung RT R Multiple bone metastasis 1.66 1.03–2.68
et al. No EGFR-targeted 2.935 1.61–5.45
[20] treatment in
adenocarcinoma
ECOG >2 1.86 1.19–2.89
Biological effective 2.148 1.3–3.54
dose<40 Gy
Other tissue than 1.74 1.12–2.70
adenocarcinoma
Kumar 7 2014 2007–2011 87 52 14 (1–120) Nasopharynx Mixed R KPS 80–100 0.07 0.03–0.18
et al. No major internal organ 0.25 0.11–0.54
[21] One vertebral metastasis 0.23 0.09–0.64
Leithner 7 2008 1998–2006 69 20 14 (N/A) Mixed Surgery R Rapid growth tumor 9.32 3.87–22.5
et al. Visceral metastases 2.17 1.15–4.09
[22]
(Continued)
694 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Table 1
(Continued)
Mean
Publication Collection No. age Median survival Tumor
Author Nos. year year of pts. (y) (95% CI) (mo) type Treatment Type Prognostic factors HR 95% CI
Mizumoto 7 2008 2002–2006 544 63 5.9 (4.9–6.8) Mixed RT R Age >70 1.433 1.14–1.80
et al. Multiple bone metastasis 1.647 1.17–2.32
[23] Elevated serum calcium 2.57 1.87–3.54
Previous CMT 1.826 1.47–2.27
ECOG 3–4 2.191 1.79–2.68
Unfavorable primary 3.326 2.62–4.22
tumor
Visceral metastases 1.694 1.36–2.11
Moon 7 2011 1987–2008 182 56 8 (6–10) Mixed Surgery R Preoperative ECOG 4 4.984 1.955–12.707
et al. Preoperative ECOG 3 3.206 1.647–6.244
[24] Tomita score 8–10 7.842 2.862–21.485
Tomita score 6–7 3.164 1.242–8.061
Nemelc 7 2014 2000–2010 81 59 9.5 (6.875–12.125) Mixed Surgery R ASIA-C versus ASIA A 13.03 4.57–37.16
et al. Multiple myeloma versus 0.22 0.05–0.95
[25] breast
Lung cancer versus 4.52 1.73–11.78
breast
Colorectal versus breast 5.53 1.60–5.53
Oh et al. 6 2012 1996–2008 52 60.6 N/A Mixed Surgery R Preoperative albumin 1.93 1.073–6.504
[26] level<3.5 g/dL
Rapid growth primary 3.15 1.276–7.081
site
Park 7 2011 2001–2008 103 54.6 10 (8.21–11.80) Mixed Surgery R No postoperative 1.586 1.021–2.645
et al. ambulation
[27] Tokuhashi score (9–11) 0.524 0.335–0.820
versus 0–8
Good prognostic cancer 0.627 0.479–0.899
Pointillart 7 2011 2005–2007 142 61.8 5 (N/A) Mixed Surgery R ASA score >3 2.273 1.049–4.927
et al. Vascular disease as a 2.461 1.347–4.497
[28] comorbidity
Preoperative 2.471 1.377–4.437
chemotherapy
Subjective pain score>6 1.988 1.153–3.428
KPS 80–100 0.978 0.964–0.992
Primary lung tumor 4.44 1.35–14.62
Weight loss 2.608 1.286–5.289
Rades 6 2006 1992–2005 1852 N/A N/A Mixed RT R Ambulator status before 0.47 0.39–0.56
et al. RT
[29] Other bone metastases at 1.29 1.04–1.59
time of RT
Interval tumor diagnosis 0.8 0.75–0.85
to MSCC >15 mo
Motor deficits time 0.72 0.68–0.77
before RT>14 d
Lung cancer 1.23 1.18–1.29
Visceral metastases at the 4.89 4.26–5.60
time of RT
Rades 6 2013 1992–2011 2029 N/A N/A Mixed RT R Non-ambulatory before 1.94 1.65–2.29
et al. RT
[30] Further bone metastases 1.34 1.12–1.61
Interval from diagnosis 1.22 1.14–1.29
to MSCC<15 mo
Time to development of 1.73 1.53–1.95
motor deficits <7 d
ECOG 3–4 1.46 1.2–1.74
Male 1.15 1.01–1.30
Lung cancer 1.18 1.03–1.23
Visceral metastases 4.46 3.92–5.08
Rades 7 2015 N/A 58 N/A N/A Head-and- RT R Non-ambulatory before 4.31 1.85–10.20
et al. Neck RT
[31] Visceral metastases 7.2 3.17–17.58
Rades 7 2012 1992–2010 171 N/A N/A Breast RT R Other bone metastases 1.93 1.18–3.13
et al. Time developing motor 1.55 1.30–1.86
[33] deficits<14 d
Non-ambulatory before 1.75 1.23–2.50
RT
Number of involved 1.27 1.01–1.60
vertebrae≥3
Visceral metastases 7.6 5.39–10.84
(Continued)
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 695

Table 1
(Continued)
Mean
Publication Collection No. age Median survival Tumor
Author Nos. year year of pts. (y) (95% CI) (mo) type Treatment Type Prognostic factors HR 95% CI
Rades 6 2014 N/A 552 N/A N/A Mixed RT R Non-ambulatory before 2.17 1.72–2.72
et al. RT
[32] Extraspinal organs≥2 1.61 1.47–1.77
sites
Interval from diagnosis 1.26 1.13–1.42
to RT<15 mo
Time developing motor 1.34 1.16–1.55
deficits<14 d
ECOG 3–4 2.41 1.98–2.92
Male 1.32 1.05–1.67
Lung and other tumor 1.09 1.04–1.13
(vs. breast and
prostate)
Rades 7 2012 N/A 121 N/A N/A Colorectal RT R Non-ambulatory before 2.8 1.78–4.37
et al. cancer RT
[34] The time of developing 1.33 1.01–1.77
motor deficits<14 d
ECOG 3–4 2.96 1.84–4.74
Visceral metastases 3.19 1.94–5.47
Rades 6 2011 N/A 382 N/A N/A Mixed RT R Non-ambulatory before 2.47 1.46–4.05
et al. RT
[35] ECOG 3–4 2.37 1.48–3.73
Low-dose RT (30 Gy/10 1.64 1.11–2.44
fractions)
Lung and other tumor 3.93 2.11–7.18
(vs. breast and
prostate)
Visceral metastases 5.4 2.83–9.88
Rades 7 2012 1992–2010 356 N/A N/A Lung RT R Ambulatory before RT 0.58 0.42–0.79
et al. Bone metastases 1.38 1.08–1.76
[36] Cancer diagnosis interval 0.84 0.70–0.99
to RT >15 mo
Time developing motor 0.78 0.69–0.88
deficits>7 d
ECOG performance 1.45 1.03–2.03
score 3–4
Male 1.32 1.01–1.75
Visceral metastases 2.87 2.17–3.85
Rades 7 2012 1992–2010 436 N/A N/A Prostate RT R Non-ambulatory before 3.17 2.16–4.69
et al. RT
[37] Other bone metastases 1.53 1.04–2.28
Interval from diagnosis 1.27 1.05–1.53
MSCC to RT <15 mo
ECOG 3–4 2.67 1.76–4.12
Visceral metastases 4.15 2.76–6.15
Rades 7 2014 1998–2011 69 N/A N/A Renal cell RT R Non-ambulatory before 3.03 1.51–6.13
et al. carcinoma RT
[38] Number of involved 2.65 1.64–4.52
extraspinal organs≥2
Interval from diagnosis 1.46 1.08–1.99
of MSCC to RT <15
mo
Rades 7 2008 1995–2007 164 N/A 7 (N/A) Mixed RT R Non-ambulatory before 3.25 1.80–5.85
et al. recurrence RT
[57] Time to develop motor 1.54 1.02–2.20
deficits<14 d
ECOG 3–4 4.75 2.59–8.78
Visceral metastases 15.44 7.59–39.73
Rades 6 2011 2006–2007 265 N/A N/A Mixed RT P No bisphosphonates after 2.86 1.82–4.79
et al. radiotherapy
[54] ECOG 3–4 2.09 1.47–2.99
Number of involved 111 1.01–1.23
vertebrae≥4
Visceral metastases 1.66 1.23–2.24
Non-ambulatory before 1.14 1.02–1.92
RT
Rief 8 2014 200–2012 303 64.1 9.8 (N/A) NSCL RT R Bone metastases>1 site 1.09 1.03–1.15
et al.
[39]
Sciubba 8 2007 1993–2001 87 53 21 (12–27) Breast Surgery R Estrogen receptor 3.7 1.7–8.1
et al. negative
[40]
(Continued)
696 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Table 1
(Continued)
Mean
Publication Collection No. age Median survival Tumor
Author Nos. year year of pts. (y) (95% CI) (mo) type Treatment Type Prognostic factors HR 95% CI
Sellin 7 2015 2005–2013 37 62.2 16.3 (7.4–25.3) RCC RT R Metastasis<12 mo after 2.57 1.12–5.90
et al. primary diagnosis
[41] Local progression 3.69 1.64–8.29
KPS score <70 4 1.67–10.0
Progression of systemic 13.15 4.3–40.19
disease
Switlyk 7 2015 2007–2008 317 65 8.2 (0.4–80) Mixed RT R Albumin <30 g/L 2.9 1.5–5.6
et al. Strong opioids use 2.7 1.4–3.7
[6] KPS 50–70 2.3 1.5–3.5
KPS 10–40 2.6 1.0–6.4
Prostate versus breast 2.2 1.2–3.9
Other versus breast 6.2 3.6–10.7
Visceral metastases 2 or 1.6 1.0–2.6
more locations
Tabouret 7 2015 2004–2010 148 60 8.9 (4.8–13) Mixed Surgery R ASA 3–4 3.07 1.78–5.31
et al. Extra-bone multiple 2.41 1.48–3.94
[42] systemic metastases
KPS>70 0.5 0.3–0.84
Tatsui 8 2014 1993–2007 267 59.2 11 (9.5–13) RCC Surgery R Progressing systemic 4.1 2.9–5.8
et al. disease at spine
[58] surgery
Preoperative neurologic 1.8 1.2–2.7
deficit
Fuhrman grade 4 1.8 1.2–2.6
Wang 8 2014 1992–2011 151 58.7 21.2 (15.9–25.1) Breast Surgery R Estrogen receptor 0.52 0.329–0.92
et al. positive
[43] Progesterone receptor 0.41 0.19–0.88
positive
Hormonal receptor 0.48 0.27–0.84
positive
Ratasvuori 8 2014 1999–2009 307 62 12.9 (N/A) Breast Surgery P Age over 65 1.398 1.083–1.804
et al. Multiple skeletal 2.331 1.449–3.751
[55] metastases
146 73 6.1 (N/A) Prostate Surgery P Time from primary 2.107 1.227–3.617
diagnosis to
operation>6 mo
122 65 12.1 (N/A) RCC Surgery P Operation other than en 1.833 1.032–3.255
bloc resection
Pelvis metastases (vs. 0.4 0.178–0.899
lower limb
metastases)
307 62 12.9 (N/A) BREAST SURGERY P KPS<70 1.506 1.171–1.936
97 64 3.6 (N/A) LUNG SURGERY P KPS<70 1.715 1.119–2.628
122 65 12.1 (N/A) RCC SURGERY P KPS<70 2.955 1.894–4.61
97 64 3.6 (N/A) Lung SURGERY P Presence of organ 2.049 1.299–3.233
metastases
146 73 3.1 (N/A) Prostate SURGERY P Presence of organ 1.911 1.216–3.002
metastases
Weber 7 2013 N/A 95 N/A N/A Prostate RT R Non-ambulatory before 2.7 1.46–5.24
et al. RT
[44] Number of involved 1.88 1.09–3.14
extraspinal organs>1
Cancer diagnosis to RT 1.64 1.24–2.19
interval <15 mo
ECOG 3–4 4.69 2.04–12.28
Yamashita 6 2011 2006–2008 85 60.3 11.6 (6–17.3) Mixed Mixed P KPS 50–70 2.92 1.55–5.48
et al. Preoperative EBRT 2.14 1.12–4.10
[56] Lung cancer (vs. thyroid, 4.25 2.06–8.78
breast, etc.)
Renal cell cancer (vs. 2.6 1.13–5398
thyroid, breast, etc.)
Unremovable major 4.44 2.35–8.41
organ metastases
Yang 7 2012 2001–2009 217 55.5 6 (4.9–7.1) Mixed Surgery R Non-ambulatory status 5.397 3.56–8.181
et al. Preoperative Tomita 1.439 1.003–2.065
[5] score>6 points
Rapid growth tumor 1.607 1.046–2.469
Yasuda 8 2013 1978–2010 45 N/A 27.2 (N/A) RCC Surgery R No zoledronic acid 1.89 1.27–2.96
et al. treatment
[45] Calcium (>10 mg/dL) 2.58 1.31–4.69
(Continued)
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 697

Table 1
(Continued)
Mean
Publication Collection No. age Median survival Tumor
Author Nos. year year of pts. (y) (95% CI) (mo) type Treatment Type Prognostic factors HR 95% CI
Yeung 6 2014 2001–2011 128 60.2 3.1 (N/A) Mixed Mixed R KPS 10–40 6.89 3.07–15.46
et al. KPS 50–70 2.62 1.41–4.86
[46] Lung, bladder, stomach 1.86 1.06–3.26
versus breast, prostate
Liver, unidentified versus 2.49 1.23–5.01
breast, prostate
Kidney, uterus 2.76 1.15–6.65
Unremovable visceral 1.73 1.15–2.62
metastasis
Removable visceral 4.82 1.11–20.99
metastasis
Zhang 8 2013 2003–2011 36 49.9 26.3 HCC Surgery R Tomita score of 8 or 3.549 1.374–9.168
et al. more
[47]
Bollen 6 2014 2001–2010 1043 64.8 8.9 (7.4–10.3) Mixed Mixed R KPS 10–70 1.9 1.6–2.2
et al. Rapid growth tumor 3.5 2.9–4.4
[48] Moderate growth tumor 1.6 1.3–2.1
Visceral metastases 1.5 1.3–1.7
Rades 6 2015 N/A 142 N/A N/A Mixed RT R ECOG 3–4 2.76 1.69–4.39
et al. Non-ambulatory before 2.49 1.36–4.26
[49] RT
Visceral metastases at the 4.47 2.44–7.70
time of RT
Rades 7 2013 1995–2011 510 N/A N/A Breast RT R Non-ambulatory before 2.13 1.44–3.13
et al. RT
[50] Other bone metastases 2.51 1.19–5.26
Cancer diagnosis interval 1.29 1.03–1.61
to RT<15 mo
Time of developing 1.45 1.14–1.85
motor deficits<7 d
ECOG 3–4 2 1.34–3.00
Shorter-course 1.25 1.03–1.51
radiotherapy (1×8 or
5×4 Gy)
Visceral metastases 11.04 6.81–18.43
Rades 7 2014 N/A 131 N/A N/A Non–small RT R Non-ambulatory before 2.07 1.32–3.26
et al. cell lung RT
[51] cancer Number of involved 1.6 1.28–2.00
extraspinal organs >1
Time to developing 1.31 1.01–1.72
motor deficits<14 d
ECOG 3–4 2.36 1.39–4.11
Female 1.67 1.06–2.73
Non-adenocarcinoma 1.37 1.07–1.76
histology
Weber 7 2014 N/A 145 N/A N/A Breast RT R Non-ambulatory before 2.32 1.34–5.32
et al. RT
[53] Number of involved 2.19 1.61–3.00
extraspinal organs>1
Time to developing 1.5 1.05–2.17
motor deficits<7 d
ECOG 3–4 4.09 2.17–8.14
KPS, Karnofsky Performance Score; LDH, lactate dehydrogenase; ECOG, Eastern Cooperative Oncology Group; HCC, hepatocellular carcinoma; RT, radiotherapy; RCC, renal
cell carcinoma; EGFR, epidermal growth factor receptor; CMT, chemotherapy; ASIA, American Spinal Injury Association; ASA, American Society of Anesthesiologists; MSCC,
metastatic spinal cord compression tumor; NSCL, non–small cell lung cancer.

deficit before surgery, non-ambulatory status before RT,
non-ambulatory status before surgery, presence of bone
metastases, presence of multiple bone metastases (>2 sites),
presence of multiple spinal metastases (>3 sites), develop-
ment of motor deficit in <7 days before initiating RT,
development of motor deficit in <14 days before initiating
RT, time interval from cancer diagnosis to RT <15 months,
KPS 10–40, KPS 50–70, KPS<70, ECOG grade 3–4, male
gender, presence of visceral metastases, moderate growth
tumor on TS classification, and rapid growth tumor on TS
classification.
The 17 poor prognostic factors with similar variables that
were mentioned in at least three studies were pooled in the
meta-analysis. The details of meta-analysis results are shown
in Table 5.
The pooled HR of all factors showed statistical signifi-
cance except preoperative non-ambulatory status (p=.377). The
sensitivity analysis of all factors demonstrated that exclu-
sion of each study did not affect the conclusion of pooled HRs
except the preoperative neurologic deficit. The pooled HR of
preoperative neurologic deficit resulted in no statistical sig-
nificance if Hosono et al. (p=.128) [18] or Tatsui et al. (p=.125)
698 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
Table 2
Present specific and general poor prognostic factors of each primary tumor site
Tumor Specific factors
Breast cancer • Negative estrogen receptor [40,43]
• Negative progesterone receptor [43]
• Negative hormonal receptor [43]
• Poorly or undifferentiated pathology [11]
• shorter course radiotherapy (1×8 Gy or
5×4 Gy) [50]
Colorectal cancer
Hepatocellular carcinoma • Biologically effective dose Gy10<38 [13]
(HCC) • Uncontrolled primary HCC [13]
Head and neck cancer
Nasopharyngeal cancer
Lung cancer • biological effective dose<40 Gy [20]
• other tissue than adenocarcinoma [20]
• No use of EGFR-targeted treatment in
adenocarcinoma [20]
Non–small cell lung cancer Non-adenocarcinoma pathology [51]
Prostate cancer
General factors
• Age more than 65 years [55]
• Non-ambulatory status before radiotherapy [33,50,53]
• Multiple extraspinal bone metastases [33,50,53,55]
• Admission through emergency room [11]
• Increasing time from diagnosis of cancer to surgery (per 1 y) [11]
• Interval from cancer diagnosis to radiotherapy of MSCC<15 mo [50]
• Time of developing motor deficits
o <7 days [50]
o <14 days [33,53]
• KPS<70 [55]
• ECOG performance status 3–4 [50,53]
• Number of involved vertebrae≥3 [33]
• Visceral metastases [33,50]
• Non-ambulatory before RT [34]
• Time of developing motor deficits<14 d [34]
• ECOG performance status 3–4 [34]
• Visceral metastases [34]
• Albumin<37 g/dL [12]
• Extrahepatic metastases other than bone [13]
• LDH>200 U/L [12]
• ECOG performance status 3–4 [13]
• Tomita score of 8 or more [47]
• Non-ambulatory status before radiotherapy [31]
• Visceral metastases [31]
• Poor KPS [21]
• Major internal organ metastases [21]
• Multiple vertebral metastases [21]
• Non-ambulatory before RT [36]
• Multiple bone metastasis [20]
• Other bone metastases [36]
• Interval from cancer diagnosis to RT<15 mo [36]
• Time developing motor deficits<7 d [36]
• ECOG ≥2 [20]
• ECOG 3–4 [36]
• KPS<70 [55]
• Male [36]
• Visceral metastases [36,55]
• Non-ambulatory status before RT [51]
• Multiple bone metastases [39,51]
• Time to developing motor deficits <14 d [51]
• Male [51]
• Non-ambulatory before RT [37] [44]
• Other bone metastases [37]
• Multiple extraspinal metastases [44]
• Interval from cancer diagnosis MSCC to RT<15 mo [37]
• Time from primary diagnosis to operation<6 mo [55]
• Interval from cancer diagnosis to RT<15 mo [44]
• ECOG performance score 3–4 [37,44]
• Visceral metastases [37,55]
• KPS 50–70 [15]
(Continued)
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 699

Table 2
(Continued)
Tumor Specific factors General factors
Renal cell carcinoma • No zoledronic acid treatment [45] • Non-ambulatory before RT [38]
• Fuhrman grade 4 [58] • Number of involved extraspinal organs ≥2 [38]
• Serum calcium >10 mg/dL [45]
• Progressing systemic disease at spine surgery [58]
• Interval from cancer diagnosis to RT ≤15 mo [38]
• Diagnosis of metastasis less than 12 mo after primary diagnosis [41]
• Local disease progression after spinal SRS [41]
• Preoperative neurologic deficit [58]
• Operation strategy other than en bloc resection [55]
• Pelvis metastases compared with lower limb metastases [55]
• KPS <70 [41,55]
• Progression of systemic disease at time of spinal SRS [41]
• Revised Tokuhashi score 4–9 compared with 10–12 [17]
Recurrence of tumor in the • Non-ambulatory before RT [57]
previously irradiated part • Time to development of motor deficits before RT<14 d [57]
of spine • ECOG performance score 3–4 [57]
• Visceral metastases [57]
ECOG, Eastern Cooperative Oncology Group; RT, radiotherapy; MSCC, metastatic spinal cord compression tumor; LDH, lactate dehydrogenase; EGFR,
epidermal growth factor receptor; KPS, Karnofsky Performance Score; CMT, chemotherapy; HCC, hepatocellular carcinoma; SRS, stereotactic radiosurgery.

Table 3
Present poor prognostic factors from studies that are reported in various primary tumor sites
Factor Variables
Age Age >70 [23]
Systemic disease Progression of systemic disease at time of spinal SRS [6]
multiple systemic metastases [42], weight loss [28]
Non-ambulatory before RT Non-ambulatory before RT [30,32,35,49,54]
Preoperative non-ambulatory Preoperative non-ambulatory [5,9]
Postoperative non-ambulatory Postoperative non-ambulatory [27]
ASA status ASA score 3–4 [28,42]
Extraspinal bone metastases Presence of extraspinal bone metastases [19,29,30]
Multiple extraspinal bone metastases (≥2 sites) [23,32]
Number of spinal metastases Multiple spinal metastases ≥2 [16], ≥3 [14], ≥4 [54]
Abnormal blood test Elevated serum calcium [23], preoperative albumin level<3.5 g/dL [26]
Comorbidities Charlson comorbidity index ≥2 [9], vascular disease as comorbidity [28]
Previous chemotherapy Previous chemotherapy before RT [23], preoperative chemotherapy [28]
Preoperative radiotherapy Preoperative EBRT [56]
Interval from diagnosis to spinal metastases Disease free before developing spinal metastases<12 mo [19]
Interval from diagnosis to MSCC<15 mo [29,30]
Time developing motor deficits before RT Time developing motor deficits before RT<7 d [30]
Time developing motor deficits before RT<14 d [29,32]
Interval from diagnosis to RT Interval from cancer diagnosis to RT≤15 mo [32]
Preoperative neurologic deficit Preoperative paresis [18], ASIA-C compared with ASIA-E [25]
Severe pain Strong opioids use [6], present pain [18], subjective pain score>6 [28]
Karnofsky Performance Score (KPS) [3] KPS 10–40 [6,10,16,46], KPS 50–70 [6,16,46,56], KPS<70 [28,42,48]
Eastern Cooperative Oncology Group (ECOG) ECOG 3–4 [23,30,32,35,54], preoperative ECOG status 3–4 [24]
No adjuvant therapy No postoperative adjuvant therapy [14], no administration of bisphosphonates
after radiotherapy [54]
Radiotherapy course Low-dose RT (30 Gy/10 fractions) [35]
Revised Tokuhashi score RTS 9–11 [27]
Tomita score Preoperative Tomita score>6 [5]
Tomita score 6–7 compared with Tomita score 2–3 [24]
Tomita score 8–10 compared with Tomita score 2–3 [24]
Gender Male [30,32]
Visceral metastases [3,10,19,22,23,29,30,35,48,49,52,54] Treatable visceral metastases [16,46]
Non-treatable visceral metastases [16,46,56]
Visceral metastases≥2 locations [6]
ECOG, Eastern Cooperative Oncology Group; RT, radiotherapy; KPS, Karnofsky Performance Score; ASIA, American Spinal Injury Association; SRS,
stereotactic radiosurgery; RTS, revised Tokuhashi score; EBRT, external beam radiotherapy; MSCC, metastatic spinal cord compression tumor; ASA, Amer-
ican Society of Anesthesiologists.
700 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Table 4
Present prognostic factor comparing primary tumor site
Poor prognosis Good prognosis
Primary tumor Radio-resistant tumor [9] Radiosensitive tumor
site [3,52] Moderate growth tumor of Tomita classification [19,48] Slow growth tumor of Tomita classification
Fast growth tumor of Tomita classification [5,10,16,19,22,48] Slow growth tumor of Tomita classification
Lung, bladder, stomach [46] Breast, prostate, thyroid
Liver, unidentified [46] Breast, prostate, thyroid
Renal cell, uterus [46] Breast, prostate, thyroid
Renal cell [56] Breast, prostate, thyroid
Poor prognosis tumor (other than good prognosis group) [18,23,26,27,35] Myeloma, thyroid, renal cell, breast, prostate, lymphoma
Renal cell [56] Thyroid
Prostate [6], lung, colorectal cancer [25], other tumor except breast [6] Breast
Lung [28–30] Other tumor except lung
Lung [32] Breast, prostate
Lung [56] Breast, prostate, thyroid
Slow growth=breast, prostate, etc.; moderate growth=renal cell, uterus, etc.; and rapid growth=lung, stomach, etc.

Fig. 2. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for Karnofsky Performance Score.
Table 5
Show results of meta-analysis including pooled HR, 95% CI, sensitivity analysis, and publication bias

Pooled Heterogeneity Sensitivity Affected Publication bias

Prognostic factors N Total pts. HR range Pooled HR 95% CI (I2) Model p analysis study Begg Egger
Non-ambulatory status before RT 17 7,478 1.14–4.31 2.15 1.89–2.44 48.4% Random <.0001 No effect None 0.021 0.086
[29–38,44,49–51,53,54,57]

P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Preoperative non-ambulatory [5,9,14,15,18] 5 755 0.90–2.36 1.20 0.80–1.82 63.6% Random .377 No effect None 1 0.618
Presence of bone metastases 10 5,879 0.95–2.51 1.41 1.27–1.56 0% Fixed <.0001 No effect None 0.245 0.154
[13,19,22,29,30,33,34,36,37,50]
Multiple bone metastases (>2 sites) 10 2,426 1.09–2.65 1.78 1.43–2.20 90.7% Random <.0001 No effect None 0.929 0.011
[20,32,36,38,39,42,44,53,55] 9 2,123 1.60–2.65 1.70 1.57–1.83 26.7% Fixed <.0001 No effect None 0.061 0.019
Spinal metastases ≥3 sites [14,29,32,33,36,38,44] 8 3,710 0.99–1.94 1.13 1.02–1.24 54% Random .014 No effect None 0.026 0.002
Time developing motor deficits before RT<7 d 6 3,167 1.22–1.90 1.48 1.26–1.74 60.9% Random <.0001 No effect None 0.573 0.993
[30,31,36,38,51,53]
Time developing motor deficits before RT <14 d 7 3,086 1.31–1.55 1.39 1.32–1.47 0% Fixed <.0001 No effect None 0.453 0.739
[29,32–34,44,51,57]
Interval from cancer diagnosis to RT <15 mo 9 6,030 1.19–1.64 1.25 1.20–1.30 0% Fixed <.0001 No effect None 0.012 0.055
[29,30,32,36–38,44,50,51]
Preoperative neurologic deficit [18,25,58] 3 513 1.79–13.03 2.97 1.28–6.91 84.2% Random .012 Effect Hosono et al. 0.117 0.153
[18], Tatsui
et al. [58]
KPS 10–40 [6,10,16,46] 4 747 2.60–6.89 3.17 2.26–4.47 30.3% Fixed <.0001 No effect None 1 0.666
KPS 50–70 [10,15,16,22,46,56] 7 922 1.30–3.97 2.09 1.66–2.63 30.6% Fixed <.0001 No effect None 0.348 0.451
KPS<70 [12,14,21,28,41,42,48,55] 8 2,129 0.75–14.39 1.89 1.33–2.67 93.7% Random .0003 No effect None 0.788 0.011
[12,14,21,28,41,42,48,55] 7 1,987 0.75–14.39 2.05 1.49–2.82 79.1% Random <.0001 No effect None 0.532 0.580
ECOG 3–4 [13,20,23,29–32,34–38,44,49–51,53,54,57] 19 8,175 1.18–4.75 2.14 1.83–2.50 67.7% Random <.0001 No effect None 0.152 0.087
Male [5,24,29,30,32,36,39,46,51] 9 5,750 1.17–1.67 1.19 1.10–1.27 0% Fixed <.0001 No effect None 0.144 0.013
Presence of visceral metastases 21 9,026 1.09–15.44 3.10 2.35–4.07 94.2% Random <.0001 No effect None 0.272 0.922
[9,10,14,19,21–23,29–31,33–37,48–50,54,55,57]
Moderate growth tumor of Tomita [5,10,16,19,22,48] 6 1,644 1.07–1.80 1.55 1.30–1.85 29.9% Fixed <.0001 No effect None 0.851 0.944
Fast growth tumor of Tomita [5,10,16,19,22] 5 731 1.70–9.32 3.06 1.65–5.68 85% Random .0004 No effect None 0.327 0.246
RT, radiotherapy; KPS, Karnofsky Performance Score; ECOG, Eastern Cooperative Oncology Group.

701
702 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Fig. 3. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for bone and spinal metastases.

[58] was omitted. Additionally, omitting the study of Rief et al.
[39] for multiple bone metastases and the study of Pointillart
et al. [28] for KPN<70 decreased I2 to less than 90%. There
was no significant publication bias. The forest plot and funnel
plot are shown in Figs. 2–7.
Discussion
This systematic review included 51 publications for qual-
itative study and 43 publications for meta-analysis. There were
13 poor prognostic factors specific to each primary cancer
and 25 poor prognostic factors that applied to all groups. The
HRs of 17 factors mentioned in at least three publications were
pooled. The sensitivity analysis of all factors revealed that
pooled estimate of effect did not change when each study was
omitted. The homogeneity analysis showed large homoge-
neity. Most of the included studies were single-center,
retrospective studies.
The RTS was the most popular prognostic scoring system.
However, this system had overall predictive value of 66% [65].
Lee et al. reported that the RTS was slightly more accurate
than TS, but both showed low accuracy in predicting 6 months’
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 703

Fig. 4. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for interval from cancer diagnosis and time developed motor deficit
before radiotherapy.

residual survival [66]. Yu et al. reported that only 8.6% of
patients with lung cancer spinal metastasis actually fol-
lowed the RTS. Furthermore, RTS could predict survival
accurately only in those patients with visceral tumors and
disease progression after first-time chemotherapy, simulta-
neously. They suggested that both factors should be added
into the RTS system to increase validity and accuracy [67].
However, our study found new factors that were not men-
tioned in RTS, including gender, ambulatory status, hormonal
receptor, age, interval between cancer diagnosis and RT, ab-
normal blood test (calcium, albumin, and lactate
dehydrogenase), time to development of motor deficit, ASA
score, comorbidities, disease-free interval, severe pain, ad-
juvant therapy, and RT course. Additionally, the effect of each
general prognostic factor had great variance between primary
tumor sites. For example, the HRs of positive visceral me-
tastases in prostate, breast, and lung cancer were 1.91–4.15,
7.6–11.04, and 2.04–2.87, respectively. The HR of studies that
report with mixed primary tumor ranged from 1.08 to 4.89.
We hypothesized that if RTS added and accurately weighted
these significant factors, the accuracy of RTS could in-
crease significantly.
Moreover, each metastatic cancer had specific prognos-
tic factors that may be related to the response rate of
adjuvant therapy. Specific prognostic factors for metastatic
breast cancer included hormonal receptor, poorly differenti-
ated pathology, and a course of RT. Specific prognostic
factors for metastatic lung cancer were non-adenocarcinoma
704 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708

Fig. 5. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for non-ambulatory status and neurologic deficit.

type and no epidermal growth factor receptor-targeted therapy
in adenocarcinoma type. Higher prostate-specific antigen
level was the only specific factor in prostate metastatic
cancer. Unfortunately, current prognostic scores did not add
any specific factors in their systems. Currently, there are
some new prognostic scores that are specific to primary
tumor including prostate cancer [15] and nasopharyngeal
cancer [21]. However, these new scoring systems do not
include all potential nonspecific prognostic factors and lack
external validity studies.
Laufer et al. reported a new approach to spinal meta-
static tumors termed the “NOMS” approach. This approach
consists of neurologic, oncologic, mechanical, and system-
ic considerations. Laufer et al.’s study mentions that “systemic
disease assessment determines what a patient can tolerate
physiologically and is dependent on extent of tumor dissem-
ination, medical comorbidities and tumor histology” [68].
However, there are no clearly identified specific medical
morbidities. Our study found many systemic factors that
may be used for determining prognosis and that may guide
the treatment in NOMS approach including ASA status,
Charlson comorbidity index >2, vascular disease, elevated
serum calcium, preoperative albumin level<3.5 g/dL, severe
pain, and weight loss.
There are several limitations to this study. First, most of
the included studies were retrospective studies that did not
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708 705

Fig. 6. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for primary tumor and visceral metastases.
706 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
Fig. 7. Forest plot showing pooled hazard ratio and funnel plot showing publication bias for male and Eastern Cooperative Oncology Group grade.
have a specific survival period (eg 5-year or 10-year overall
survival). Second, there were many studies reported by
Rades et al. that did not clearly define the source of
enrolled patients, median survival, and were conducted
only on radiated patients. Third, 75% (38 of 51) of included
studies had a Newcastle-Ottawa Scale score of 6–7, which
indicates only moderate quality. Finally, we selected only
English publications, which may have some selection bias.
A multicenter prospective cohort study is needed to prove
the effect of these independent prognostic factors.
Conclusion
This systematic review and meta-analysis showed that in-
dependent poor prognostic factors can be categorized into
tumor-specific and nonspecific factors. The current prognos-
tic scoring system has moderate accuracy. A tumor-based
prognostic scoring system that combines all specific and
general factors of each primary tumor may enhance the ac-
curacy of survival prediction.
References
[1] Tokuhashi Y, Matsuzaki H, Oda H, Oshima M, Ryu J. A revised scoring
system for preoperative evaluation of metastatic spine tumor prognosis.
Spine 2005;30:2186–91.
[2] Tomita K, Kawahara N, Kobayashi T, Yoshida A, Murakami H,
Akamaru T. Surgical strategy for spinal metastases. Spine 2001;26:298–
306.
[3] van der Linden YM, Dijkstra SP, Vonk EJ, Marijnen CA, Leer JW;
Dutch Bone Metastasis Study Group. Prediction of survival in patients
with metastases in the spinal column: results based on a randomized
trial of radiotherapy. Cancer 2005;103:320–8.
[4] Bauer HC, Wedin R. Survival after surgery for spinal and extremity
metastases. Prognostication in 241 patients. Acta Orthop Scand
1995;66:143–6.
[5] Yang SB, Cho W, Chang UK. Analysis of prognostic factors relating
to postoperative survival in spinal metastases. J Korean Neurosurg Soc
2012;51:127–34.
[6] Switlyk MD, Kongsgaard U, Skjeldal S, et al. Prognostic factors in
patients with symptomatic spinal metastases and normal neurological
function. Clin Oncol 2015;27:213–21.
[7] Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred
reporting items for systematic reviews and meta-analyses: the PRISMA
statement. Int J Surg 2010;8:336–41.
 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
[8] Wells G, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale
(NOS) for assessing the quality of nonrandomised studies in meta-
analyses. Available at: http://www.ohri.ca/programs/clinical
_epidemiology/oxford.asp. Accessed December 30, 2015.
[9] Arrigo RT, Kalanithi P, Cheng I, et al. Predictors of survival after
surgical treatment of spinal metastasis. Neurosurgery 2011;68:674–81
discussion 81.
[10] Bollen L, de Ruiter GC, Pondaag W, et al. Risk factors for survival
of 106 surgically treated patients with symptomatic spinal epidural
metastases. Eur Spine J 2013;22:1408–16.
[11] Cahill KS, Chi JH, Day AL, Claus EB. Trends in survival after surgery
for breast cancer metastatic to the brain and spinal column in medicare
patients: a population-based analysis. Neurosurgery 2011;68:705–13
discussion 13.
[12] Chen H, Xiao J, Yang X, Zhang F, Yuan W. Preoperative scoring
systems and prognostic factors for patients with spinal metastases from
hepatocellular carcinoma. Spine 2010;35:E1339–46.
[13] Choi C, Seong J. Predictive factors of palliative radiotherapy response
and survival in patients with spinal metastases from hepatocellular
carcinoma. Gut Liver 2015;9:94–102.
[14] Chong S, Shin SH, Yoo H, et al. Single-stage posterior decompression
and stabilization for metastasis of the thoracic spine: prognostic factors
for functional outcome and patients’ survival. Spine J 2012;12:1083–92.
[15] Crnalic S, Lofvenberg R, Bergh A, Widmark A, Hildingsson C.
Predicting survival for surgery of metastatic spinal cord compression
in prostate cancer: a new score. Spine 2012;37:2168–76.
[16] Dardic M, Wibmer C, Berghold A, Stadlmueller L, Froehlich EV,
Leithner A. Evaluation of prognostic scoring systems for spinal
metastases in 196 patients treated during 2005–2010. Eur Spine J
2015;24:2133–41.
[17] Han S, Wang T, Jiang D, et al. Surgery and survival outcomes of 30
patients with neurological deficit due to clear cell renal cell carcinoma
spinal metastases. Eur Spine J 2015;24:1786–91.
[18] Hosono N, Ueda T, Tamura D, Aoki Y, Yoshikawa H. Prognostic
relevance of clinical symptoms in patients with spinal metastases. Clin
Orthop Relat Res 2005;436:196–201.
[19] Kataoka M, Kunisada T, Tanaka M, et al. Statistical analysis of
prognostic factors for survival in patients with spinal metastasis. Acta
Med Okayama 2012;66:213–19.
[20] Komatsu T, Kunieda E, Oizumi Y, Tamai Y, Akiba T. An analysis of
the survival rate after radiotherapy in lung cancer patients with bone
metastasis: is there an optimal subgroup to be treated with high-dose
radiation therapy? Neoplasma 2012;59:650–7.
[21] Kumar N, Tan JJ, Zaw AS, et al. Evaluation of scoring systems and
prognostic factors in patients with spinal metastases from
nasopharyngeal carcinoma. Spine J 2014;14:2946–53.
[22] Leithner A, Radl R, Gruber G, et al. Predictive value of seven
preoperative prognostic scoring systems for spinal metastases. Eur Spine
J 2008;17:1488–95.
[23] Mizumoto M, Harada H, Asakura H, et al. Prognostic factors and a
scoring system for survival after radiotherapy for metastases to the spinal
column: a review of 544 patients at Shizuoka Cancer Center Hospital.
Cancer 2008;113:2816–22.
[24] Moon KY, Chung CK, Jahng TA, Kim HJ, Kim CH. Postoperative
survival and ambulatory outcome in metastatic spinal tumors: prognostic
factor analysis. J Korean Neurosurg Soc 2011;50:216–23.
[25] Nemelc RM, Stadhouder A, van Royen BJ, Jiya TU. The outcome and
survival of palliative surgery in thoraco-lumbar spinal metastases:
contemporary retrospective cohort study. Eur Spine J 2014;23:
2272–8.
[26] Oh I-S, Kim S-I, Ha K-Y. Significant predictive values for the life
expectancy in patients with spinal metastasis following surgical
treatment. Eur J Orthop Surg Traumatol 2012;22:17–24.
[27] Park JH, Rhim SC, Jeon SR. Efficacy of decompression and fixation
for metastatic spinal cord compression: analysis of factors prognostic
for survival and postoperative ambulation. J Korean Neurosurg Soc
2011;50:434–40.
707
[28] Pointillart V, Vital JM, Salmi R, Diallo A, Quan GM. Survival
prognostic factors and clinical outcomes in patients with spinal
metastases. J Cancer Res Clin Oncol 2011;137:849–56.
[29] Rades D, Fehlauer F, Schulte R, et al. Prognostic factors for local control
and survival after radiotherapy of metastatic spinal cord compression.
J Clin Oncol 2006;24:3388–93.
[30] Rades D, Hueppe M, Schild SE. A score to identify patients with
metastatic spinal cord compression who may be candidates for best
supportive care. Cancer 2013;119:897–903.
[31] Rades D, Schild SE, Karstens JH, Hakim SG. Predicting survival of
patients with metastatic epidural spinal cord compression from cancer
of the head-and-neck. Anticancer Res 2015;35:385–8.
[32] Rades D, Weber A, Karstens JH, Schild SE, Bartscht T. Prognostic role
of the number of involved extraspinal organs in patients with metastatic
spinal cord compression. Clin Neurol Neurosurg 2014;118:12–15.
[33] Rades D, Douglas S, Veninga T, et al. Prognostic factors in a series
of 504 breast cancer patients with metastatic spinal cord compression.
Strahlenther Onkol 2012;188:340–5.
[34] Rades D, Douglas S, Huttenlocher S, et al. Prognostic factors and a
survival score for patients with metastatic spinal cord compression from
colorectal cancer. Strahlenther Onkol 2012;188:1114–18.
[35] Rades D, Panzner A, Rudat V, Karstens JH, Schild SE. Dose escalation
of radiotherapy for metastatic spinal cord compression (MSCC) in
patients with relatively favorable survival prognosis. Strahlenther Onkol
2011;187:729–35.
[36] Rades D, Douglas S, Veninga T, et al. Metastatic spinal cord
compression in non-small cell lung cancer patients. Prognostic factors
in a series of 356 patients. Strahlenther Onkol 2012;188:472–6.
[37] Rades D, Douglas S, Veninga T, et al. A survival score for patients
with metastatic spinal cord compression from prostate cancer.
Strahlenther Onkol 2012;188:802–6.
[38] Rades D, Weber A, Bartscht T, Bajrovic A, Karstens JH, Schild SE.
A new prognostic factor for the survival of patients with renal cell
carcinoma developing metastatic spinal cord compression. Strahlenther
Onkol 2014;190:667–70.
[39] Rief H, Muley T, Bruckner T, et al. Survival and prognostic factors
in non-small cell lung cancer patients with spinal bone metastases: a
retrospective analysis of 303 patients. Strahlenther Onkol 2014;190:59–
63.
[40] Sciubba DM, Gokaslan ZL, Suk I, et al. Positive and negative prognostic
variables for patients undergoing spine surgery for metastatic breast
disease. Eur Spine J 2007;16:1659–67.
[41] Sellin JN, Reichardt W, Bishop AJ, et al. Factors affecting survival
in 37 consecutive patients undergoing de novo stereotactic radiosurgery
for contiguous sites of vertebral body metastasis from renal cell
carcinoma. J Neurosurg Spine 2015;22:52–9.
[42] Tabouret E, Cauvin C, Fuentes S, et al. Reassessment of scoring systems
and prognostic factors for metastatic spinal cord compression. Spine
J 2015;15:944–50.
[43] Wang M, Jensen AB, Morgen SS, et al. Survival analysis of breast
cancer subtypes in patients with spinal metastases. Spine 2014;39:1620–
7.
[44] Weber A, Bartscht T, Karstens JH, Schild SE, Rades D. Survival in
patients with metastatic spinal cord compression from prostate cancer
is associated with the number of extra-spinal organs involved. Anticancer
Res 2013;33:4505–7.
[45] Yasuda Y, Fujii Y, Yuasa T, et al. Possible improvement of survival
with use of zoledronic acid in patients with bone metastases from renal
cell carcinoma. Int J Clin Oncol 2013;18:877–83.
[46] Yeung Y-N, Cheung K-K, Lam T-C, Cheng H-O, Chow Y-Y. A study
of the predictive value of the modified Tokuhashi score in metastatic
spinal tumour causing cord compression in a Southern Chinese
population. J Orthop Trauma Rehabil 2014;18:15–21.
[47] Zhang D, Xu W, Liu T, et al. Surgery and prognostic factors of patients
with epidural spinal cord compression caused by hepatocellular
carcinoma metastases: retrospective study of 36 patients in a single
center. Spine 2013;38:E1090–5.
708 P. Luksanapruksa et al. / The Spine Journal 17 (2017) 689–708
[48] Bollen L, van der Linden YM, Pondaag W, et al. Prognostic factors
associated with survival in patients with symptomatic spinal bone
metastases: a retrospective cohort study of 1,043 patients. Neuro Oncol
2014;16:991–8.
[49] Rades D, Huttenlocher S, Veninga T, et al. A matched-pair analysis
comparing 5x4 Gy and 10x3 Gy for metastatic spinal cord compression
(MSCC) in patients with favorable survival prognoses. Radiat Oncol
2015;10:90.
[50] Rades D, Douglas S, Schild SE. A validated survival score for breast
cancer patients with metastatic spinal cord compression. Strahlenther
Onkol 2013;189:41–6.
[51] Rades D, Weber A, Karstens JH, Schild SE, Bartscht T. Number of
extraspinal organs with metastases: a prognostic factor of survival in
patients with metastatic spinal cord compression (MSCC) from
non-small cell lung cancer (NSCLC). Anticancer Res 2014;34:2503–7.
[52] Tancioni F, Navarria P, Pessina F, et al. Assessment of prognostic factors
in patients with metastatic epidural spinal cord compression (MESCC)
from solid tumor after surgery plus radiotherapy: a single institution
experience. Eur Spine J 2012;21(Suppl. 1):S146–8.
[53] Weber A, Bartscht T, Karstens JH, Schild SE, Rades D. Breast cancer
patients with metastatic spinal cord compression. Number of extraspinal
organs involved by metastases influences survival. Strahlenther Onkol
2014;190:283–6.
[54] Rades D, Lange M, Veninga T, et al. Final results of a prospective study
comparing the local control of short-course and long-course radiotherapy
for metastatic spinal cord compression. Int J Radiat Oncol Biol Phys
2011;79:524–30.
[55] Ratasvuori M, Wedin R, Hansen BH, et al. Prognostic role of en-bloc
resection and late onset of bone metastasis in patients with bone-seeking
carcinomas of the kidney, breast, lung, and prostate: SSG study on 672
operated skeletal metastases. J Surg Oncol 2014;110:360–5.
[56] Yamashita T, Siemionow KB, Mroz TE, Podichetty V, Lieberman IH.
A prospective analysis of prognostic factors in patients with spinal
metastases: use of the revised Tokuhashi score. Spine 2011;36:910–17.
[57] Rades D, Rudat V, Veninga T, Stalpers LJ, Hoskin PJ, Schild SE.
Prognostic factors for functional outcome and survival after reirradiation
for in-field recurrences of metastatic spinal cord compression. Cancer
2008;113:1090–6.
[58] Tatsui CE, Suki D, Rao G, et al. Factors affecting survival in 267
consecutive patients undergoing surgery for spinal metastasis from renal
cell carcinoma. J Neurosurg Spine 2014;20:108–16.
[59] Demirci S, Kamer S, Kara G, Yildirim O, Esassolak M. Does the
prognosis of nasopharyngeal cancer differ among endemic and non-
endemic regions? Acta Otolaryngol 2011;131:852–60.
[60] Drzymalski DM, Oh WK, Werner L, Regan MM, Kantoff P, Tuli S.
Predictors of survival in patients with prostate cancer and spinal
metastasis. Presented at the 2009 Joint Spine Section Meeting. Clinical
article. J Neurosurg Spine 2010;13:789–94.
[61] Ogihara S, Seichi A, Hozumi T, et al. Prognostic factors for patients
with spinal metastases from lung cancer. Spine 2006;31:1585–90.
[62] Rades D, Fehlauer F, Veninga T, et al. Functional outcome and survival
after radiotherapy of metastatic spinal cord compression in patients with
cancer of unknown primary. Int J Radiat Oncol Biol Phys 2007;67:532–
7.
[63] Wibmer C, Leithner A, Hofmann G, et al. Survival analysis of 254
patients after manifestation of spinal metastases: evaluation of seven
preoperative scoring systems. Spine 2011;36:1977–86.
[64] Castel E, Lazennec JY, Chiras J, Enkaoua E, Saillant G. Acute spinal
cord compression due to intraspinal bleeding from a vertebral
hemangioma: two case-reports. Eur Spine J 1999;8:244–8.
[65] Quraishi NA, Manoharan SR, Arealis G, et al. Accuracy of the revised
Tokuhashi score in predicting survival in patients with metastatic spinal
cord compression (MSCC). Eur Spine J 2013;22(Suppl. 1):S21–6.
[66] Lee CH, Chung CK, Jahng TA, et al. Which one is a valuable surrogate
for predicting survival between Tomita and Tokuhashi scores in patients
with spinal metastases? A meta-analysis for diagnostic test accuracy
and individual participant data analysis. J Neurooncol 2015;123:267–75.
[67] Yu W, Tang L, Lin F, Yao Y, Shen Z. Accuracy of Tokuhashi score
system in predicting survival of lung cancer patients with vertebral
metastasis. J Neurooncol 2015;125:427–33.
[68] Laufer I, Rubin DG, Lis E, et al. The NOMS framework: approach to
the treatment of spinal metastatic tumors. Oncologist 2013;18:744–51.