A systematic review of MIS and open decompression surgery for spinal metastases in the last two decades

Abstract

Objective

The primary intention of this review being to produce an updated systematic review of the literature on published outcomes of decompressive surgery for metastatic spinal disease including metastatic spinal cord compression, using techniques of MIS and open decompressive surgery.

Methods

The authors conducted database searches of OVID MEDLINE and EMBASE identifying those studies that reported clinical outcomes, surgical techniques used along with associated complications when decompressive surgery was employed for metastatic spinal tumors. Both retrospective and prospective studies were analysed. Articles were assessed to ensure the required inclusion criteria was met. Articles were then categorised and tabulated based on the following reported outcomes: predictors of survival, predictors of ambulation or motor function, surgical technique, neurological function, and miscellaneous outcomes.

Results

2654 citations were retrieved from databases, of these 31 met the inclusion criteria. 5 studies were prospective, the remaining 26 were retrospective. Publication years ranged from 2000 to 2020. Study size ranged from 30 to 914 patients. The most common primary tumors identified were lungs, breast, prostate and renal cancers. One study ( Lo and Yang, 2017)¹³ reported that in those patients with motor deficit, survival was significantly improved when surgery was performed within 7 days of the development of motor deficit compared to situations when surgery was carried out 7 days after onset. This was the only study that showed that the timing of surgery plays a significant role w.r.t. survival following the onset of spinal cord compression symptoms. Four articles identified that a pre-operative intact motor function and or ambulatory status conferred a higher likelihood of a better post-operative outcome, not just in relation to survival but also in relation to post-operative ambulation as well as a greater tendency towards suitability for adjuvant treatment. Even for the same scoring system e.g. tokuhashi and its effectiveness in predicting survival, results from different studies varied in their outcome. The Karnofsky Performance Status (KPS) being the most commonly used tool to assess functional impairment, the Eastern Cooperative Oncology Group (ECOG) performance status being used in two studies. 23 studies identified an improvement in neurological function following surgery. The most common functional scale used to assess neurological outcome was the Frankel scale, 3 studies used the American Spinal Injury Association (ASIA) impairment scale for this purpose. Wound problems including infection and dehiscence appeared to be the most commonly reported surgical complication. (25 studies). The most commonly used surgical technique involved a posterior approach with decompression, with or without stabilisation. Less commonly employed techniques included percutaneous pedicle screw fixation associated with or without mini-decompression as well as anterior approaches involving corpectomy and instrumentation. 9 studies included in their data, the effect of radiation therapy in combination with surgery or as a comparison used as an alternative to surgery in spinal metastases.

Conclusions

We provide a systematic literature review on the outcomes of decompressive surgery for spinal metastases. We analyse survival data, motor function, neurological function, as well as the techniques of surgery used. Where appropriate complications of surgery are also highlighted. It is the authors’ intention to provide the reader with a reference text where this information is ready to hand, allowing for the consideration of means and methods to improve and optimise the standard of care in patients undergoing surgical intervention for metastatic spinal disease.

Mini-Abstract Published literature has identified multiple articles in relation to the management of spinal metastases. There is a paucity however of high quality and significant data identifying outcomes in relation to decompressive surgery for this condition. The present study is directed at producing a systematic review of the literature over the past 20 years using recognised databases. The salient parameters analysed include survival data, motor as well as neurological function. The technique of surgery used along with associated complications.

1. Introduction

Advances in diagnosis and the management of cancer, considered alongside a growing aging population, has resulted in more patients surviving often considerably longer than their original cancer diagnosis. In conjunction there has been an increased incidence of spinal metastases. Breast, prostate, lung and kidney tumors most commonly disseminate into the spine.¹⁰ Although in certain instances spinal metastases can be managed conservatively, few would argue that patients presenting with spinal cord compression are unlikely to preserve or improve neurological function without surgical intervention. Approximately one third of patients with spinal metastases are symptomatic on presentation or become symptomatic. Presenting symptoms and signs include pain, neurological deficits, and biomechanical instability requiring surgical treatment. The surgical objective being to reduce neurological deficits, achieve stability and improve pain, with the intention of improving the patient's quality of life.²⁶ Surgical intervention as a component in the management of symptomatic spinal cord compression (SSCC) cannot be underestimated, surgery in itself represents often the first and paramount step in patients presenting with motor deficits. ^28. Surgery related morbidity and life expectancy must always be weighed against each other. Thus, identification of eligible patients for palliative surgery is challenging. In patients with refractory pain, progressive neurological impairment, or imminent risk of fracture or instability, surgical treatment must be balanced against the presumed tumor response rate to radio- or chemotherapy. Further important indicative determinants are patients general health, the dissemination of metastatic disease, and the postoperative palliative treatment options available.⁴

Previous studies do exist on the effects of surgical decompression in the setting of spinal metastases. However, few reports will have provided as comprehensive an analysis as this present study. Here we have systemically reviewed the literature over the previous 20 years to detail those studies that have provided useful information in relation to parameters which provide an indication of survival prediction, predictors of motor function including ambulation, the technique of surgery used, neurological status and miscellaneous outcomes.

It is useful to note that meta-analysis of the available data was not attempted due to marked heterogeneity, differences in study designs and differences in reported variables etc. making this task not a suitable proposition with the information available.

2. Methods

2.1. Study search

A systematic review was performed using guidelines of the Preferred Items for Systematic Reviews and Meta-Analyses (PRISMA). Database searches of OVID MEDLINE, and EMBASE using the following search algorithm: (decompr∗ OR separate∗) AND (spine or spina∗) AND metasta∗ AND (surge∗ OR surgi∗). This approach resulted in 2654 citations (Fig. 1) allowing a systematic analysis of the previous 20 years, 2000 to 2020.

Fig. 1.

Prisma flow diagram identifying how articles have been selected based on inclusion criteria during systematic reviews.

2.2. Inclusion and exclusion criteria

Valid studies were those that stated the outcomes of decompression surgery for spinal metastases. Excluded from our analysis were animal, in vitro, biochemical, published articles not in the English language, book chapters, and case reports (where n < 5). Both retrospective and prospective studies were included if they had greater than or equal to 30 cases, minimum follow up of 12 months. Outcome criteria needed to include Frankel or ASIA scoring for neurology and at least the Karnofsky Performance index or any other validated Patient Reported Outcome Measure (PROM).

2.3. Data collection

The initial 2654 citations were evaluated. 530 duplicates were removed, the titles and abstracts of 2124 publications were screened. From this, 1945 citations were not agreeable with the inclusion criteria. Hence analysis of the complete text of the remaining 179 studies was implemented. This identified 31 eligible articles for in depth analysis. Relevant data from these articles was analysed: year of publication, whether retrospective or prospective study, number of patients in each study, the primary tumour histology, neurological and PROM outcome. Following analysis of the eligible articles, these were tabulated and categorised into the following headings: predictors of survival, predictors of ambulation or motor function, surgical technique, neurological function, and miscellaneous outcomes.

3. Results

3.1. Study characteristics

31 studies met the inclusion criteria. 5 were prospective studies, 26 were retrospective studies. Dates of publication ranged from the year 2000–2020. Patient numbers ranged from these 30 to 914 patients. The data analysed from these studies is described in Table 1, Table 2, Table 3, Table 4, Table 5.

Table 1.

Predictors of survival.

Authors & Year	Classification	Evidence Level	No. of Patients	Age (yrs) ∗	Surgery Type	Primary Tumor Site	Complications	Survival Data
Rustagi et al., 2020	Retrospective case series	3	96	60	Decompression surgery	Lung (28); breast (16); renal (13); thyroid (6); colon (4); melanoma (3) bladder (3); other (20)	29.16% (of which 11.4% were major); 3 deaths within 30 days (4%), 3 DVTs; 6 durotomy	59 observed deaths, median survival time estimated to be 6 months (95% CI: 5, 10). Also estimated that 25% of patients survive at least 22 months (95% CI: 12, 48).
Kobayashi et al., 2020	Retrospective	3	201	65.9	Posterior decompression alone (n = 29), posterior decompression and fix-ation with instrumentation (n = 182)	Lung (NSC with molecular targeted therapy) (13), Lung (Other lung cancer) (36), Prostate (hormone dependent) (16), Prostate (hormone independent) (12), Breast (hormone dependent) (12), Breast (hormone independent) (7), Colorectal (18), Kidney (17), Thyroid (14), Malignant lymphoma (12), Multiple myeloma (9), stomach (7), Bladder (4), Liver (4) Others (13), Unknown (7),	Complications not listed	Logistic regression analysis was used toidentify independent prognostic factors. In univariate analysis, the primary site, visceral metastasis, laboratory data, and performance score were significantly associated with survival. In multivariate analysis, the hazard ratios were 2.95 for moderate growth and 4.71 for rapid growth at the primary site, 1.53 for visceral metastasis, 2.94 for disseminated metast-asis, 3.15 for critical laboratory data, and 2.83 for performance status 3–4. Hence multiple factors had significant correlation with poor survival.
Gao et al., 2020	Retrospective	3	30	61.5	Circumferential decomp-ression of the spinal cord, tumor excision, reconstruc-tion and stabilisation of the spine.	Small Cell Lung Cancer (SCLC)	One patient with a postoperative complication, with surgical site infection. Treated successfully with debridement and antibiotic therapy.	No surgical-related peri-operative death occurred, however one patient died 1 month following surgery due to rapid progression of the primary lung cancer. Remaining 29 patients, reported substantial pain relief post-operatively.The mean VAS score dropped from 7.2 (range 4–10) preoperatively to 2.9 (range 1–5) postoperatively. 10 (34%) patients had an improvement in neurological function. Frankel grade showing 1-grade improvement in 9 (31%) cases and 2 grade improvement in 1 (3%) case. The general performance status of patients was also improved at least 1-grade of ECOG-PS in 9 (31%) patients, and the rate of patients with ECOG-PS of 0–2 increased from 52% (15 cases) to 69% (20 cases).
Hohenberger et al., 2018	Retrospective	3	94	64.1	65.9% of patients (n = 62) underwent laminectomy, 28.7% (n = 27) hemi-laminectomy, and 5.4% (n = 5) ventral vertebrect-omy. In 67.1% (n = 63), the tumor was partially removed (debulking) to decompress the spinal cord, and gross total resection was achieved in 32.9% (n = 31)	Adenocarcinoma of the prostate gland in 29.8% (n = 28) of patients, adeno-carcinoma of the breast in 10.6% (n = 10), and renal cell carcinomas in 9.6% (n = 9), lung cancer 8.5% (n = 8), cancer of unknown primary 5.3% (n = 5), malig-nant melanoma 7.4% (n = 7), GI tract tumor 7.4% (n = 7), Miscellaneous 21.4% (n = 20)	In-hospital mortality was 4.3% pulm-onary embolism (n = 2), myocardial infar-ction, and systemic progression of prim-ary disease) and surgery related morbid-ity 8.5% (n = 8). 3 patients had wound healing problems, 4 developed a spinal CSF leak, and 1 patient required re-resection due to persisting spinal com-pression	The most common preoperative symptoms were local or radiating pain (74% n = 70), sensory deficits (66%, n = 62), and motor deficits (59.6%, n = 56). Motor deficits were subdivided into patients with para-paresis (44.7%, n = 42) and mono = paresis (14.9% n = 14). Loss of sphincter control seen in 14.1% (n = 13), and urinary rete tion in 17.6% (n = 16). 5 patients were asymptomatic, but rout-ine staging showed the spinal metastatic mass. At discharge, significant improve- ment of or full recovery from sensory deficits was documented in 33% of patients and from motor deficits includ- paraplegin in 67.3%. Cauda equina syn- drome with bowel and bladder dysfunct-ion had significantly or fully resolved in 30.3%. 41.5% of patients benefited from a significant reduction of local or radiating pain at discharge. During the in-hospital stay, KPI had sig-nificantly improved between admission (median 60 (30–100)) and discharge (median 70 (40–100)) (p = 0.010). At follow-up, the median KPI had improved to 80. Univariate analysis indicated that the risk of poor outcome (KPI <70) at dis-charge was significantly higher for male patients (p = 0.001) and for patients with multiple spinal metastases (p < 0.001), pre-operative urinary retention, and loss of sphincter control.as independent predictors of poor funct- ional outcome at discharge (KPI <70, p < 0.05)
de Almeida et al., 2018	Retrospective	3	117	56 ± 12.5 yrs	The surgical planning was subtotal tumour removal and spinal column or spin-al nerve decompression, and stabilisation with pedicle screws when insta-bility was detected.	The most common primary site was the breast 25.6% (n = 30), followed by the lung 14.5% (n = 17), Prostate 10.3% (n = 12), Multiple myeloma 8.5% (n = 10), kidney 7.6% (n = 9), Lymphoma 6.8% (n = 8), colon 4.2% (n = 5) others 22.2% (n = 26)	Minimum survival of 5 days seen in 1 case due to surgical complications. (Complications not documented in the study)	Incidence of survival in 1 year was 71 patients (60.7%). Maximum survival of 7 years seen in 1 patient.Survival less than 6 months was obser-ved in 35 cases (30%), between 6 and 12 months in 11 cases (9.4%), and higher than 1 year in 71 cases (60.7%). According to the Tokuhashi scale the expected survival would be 72 cases (61.5%) < 6 months (0–8 points), 37 cases (31.6) between 6 and 12 months (9–11 points), and 8 cases (6.9%) > 1 year (12–15 points).
Vanek et al., 2015	Retrospective	3	166	62 ± 12 yrs	Procedures performed were posterior decompression only, anterior or posterior instrumented procedure, and combined instrument-ed surgery - iecemeal vertebrectomy	Renal cell cancer 30, lung 19, breast 25, haematological 45, prostate 20, rectum 9, thyroid gland 3, thymoma 3, GIT 1, gynaecological 1, carcinoid 1, others 9.	The series complication rate was 20% (34 patients). Of these, 6 suffered from postop hematoma, 4 patients deterior-ated in the Frankel scale, 13 patients suffered from wound healing complicat-ions, in 4 cases we encountered failure of instrumentation, and the remaining 7 patients suffered from any other medical complication, not affecting overall out-come. Symptomatic tumor recurrence was encountered in 10 patients (6%).	Median post-operative survival time was 16.0 months. Preoperative neuro-logical status influenced survival time significantly; the median survival was 5.1 months in Frankel A - C and 28.2 months in Frankel D - E (p < 0.001). Improvement on the Frankel scale was not correlated with the survival time. (p = 0.131). Age <65 years related to a significantly longer survival time (p = 0.046). The Tokuhashi score predicted patient's survival independently (p < 0.001). Other factors had no statistical significance.
Quraishi et al. 2013	Semi-prospective	2	201	61	All patients had decompre- ssion and stabilisation. using a Posterior approach i.e. 171 patients. (vertebrectomy - 31 cases combined anterior and posterior approaches - 18 patients, anterior approach only - 12 patients.	Breast 29, (15%), haemat- ological 28 (14%), renal 26 (13%), prostate 26 (13%), lung 23 (11%), GI 11 (5%), sarcoma 9 (4%) and others 49 -24%	Overall complication rate 19% (39/201). Most common being wound infection 15 (8%), Other complications included chest infection 8 (4%), neurological worsening 4 (2%), failure of the metal work 4 (2%) and pulmonary embolism 3 (1.5%).	Patients were divided into three groups, Group 1 (Tokuhashi score 0–8, n = 84), Group 2 (Tokuhashi score 9–11, n = 83) and Group 3 (Tokuhashi score 12–15, n = 34). There was no difference in the neurological outcome (Frankel grade) between Groups 1 and 2 (p = 0.34) or Groups 2 and 3 (p = 0.70). However, there was a significant difference between Groups 1 and 3 (p = 0.001), with Group 3 having a significantly better neurological outcome. Median survival was 93 days in Group 1, 229 days in Group 2 and 875 days in Group 3 (p = 0.001). The predictive value between the actual and predicted survival was 64% (Group 1), 64% (Group 2), and 69% (Group 3). The overall predictive value of the revised Tokuhashi score using Cox regression for all groups was 66%.
Jansson et al., 2006	prospective	2	282	66	Posterior decompression and stabilisation was the primary treatment and performed in 212 patients. In 47 patients the spinal cord or cauda equina was dec-ompressed only. In 23 anterior decompressions with reconstruction of the vertebral body was applied. None had a combined proc- edure, i.e. both anterior and posterior stabilisation.	Prostate 114 (40%), breast 41 (15%), renal 23 (8%), lung 19 (7%), myeloma 15 (5%), colon 14 (5%), genito-urinary 12 (4%), unknown primary 11 (4%), melanoma 8 (3%), sarcoma 6 (2%), lymphoma 3 (1%), thyroid 2 (1%), others 13 (5%)	60 complications were record- ed in 56 (20%) of the 282 patients. Systemic complications in 13 these often being associated with early death. 49 patients had local complicat- ions. Wound infections in 34, nine required re-operation with wound revision. Two infections presented late, one after 3 years in a myeloma patient and the other after 7 years in a lymphoma patient. Two of 19 patients who had a thoracotomy developed pleural exudates needing drainage. Two patients with epidural compression at T3 and T4, respectively, had wrong level dec ompression and were reoperated the next day. One remained paraplegic, the other regained normal motor function and survived 6 years.	The survival rate was 0.63 at 3 months, 0.47 at 6 months, 0.30 at 1 year, and 0.16 at 2 years. Twelve of 255 (5%) patients with motor deficits were worsened postoperatively, whereas 179 (70%) improved at least one Frankel grade. The ability to walk postoperatively was retained during follow-up in more than 80% of the patients. This review strongly advocates surgery as a means of improving function. However, complication rates were high many patients died due to their disease within months of surgery.

Table 2.

Predictors of ambulatory status or motor function.

Authors & Year	Classification	Evidence Level	No. of Patients	Age (yrs) ∗	Surgery Type	Primary Tumor Site		Complications	Outcome Scale	Outcomes
Zhu et al. 2020	Retrospective review	3	154	TOS group 54.10 MISS group 53.84	Separation surgery performed in a postero- lateral approach MISS under C-arm guidance	Breast (19%), Lung (18%) Kidney (10%), Liver (9%) Throid (4.5%) Colrectal (3%), Unknown (8%) Prostate (4%), Nasopharynx (5%), Other (15%)	Myeloma (4.5%)	Dural tear in one of 49 patients in the MISS group (2.04%) and two of 105 patients (1.90%) in the TOS group. Only one patient (2.04%) in the MISS group had a surgical wound infection, whereas six patients (5.71%) in the TOS group got infected.	FRANKEL	Pre-operative Frankel grades of patients in the two groups exhibited no significant difference (P = 0,59). Twenty three of 49 (47%) patients in MISS group and 43 of 105 (41%) patients in TOS group obtained improvement in neurological function (P = 0.48). 49% (24/49) and 55.3% (58/105) of patients had stable functional status after surgery in the MISS and TOS group, respectively (P = 0.47). Only two (4%) patients in MISS group and four (3.7%) in TOS group experienced worse postoperative neurological status.
Colangeli et al. 2020	Retrospective review	3	52	59.3	percutaneous pedicle screw fixations (PPSF)coup- led with mini-decompressi ions if required by spinal cord compressions.	Breast (27%), Renal (17%), Lung (8%), Prostate (2%), Other (44%)		3 minor and 3 major complications were reported. In one case an aseptic screw mobilization was detected without clinical consequences, while two patients had wound dehiscence healed without additional surgeries. As to the major complications, in two cases an intra- operative lesion of the dural sac during mini-decompression procedure was reported without consequences, in another patient a hematoma was reported with neurological compression symptoms, which required further surgical decompression resulting in symptom regression.	FRANKEL and VAS	Frankel were E in 37 patients, D3 in in 6 patients, D2 in 3 cases, D1 in 3 cases, B in one patient, and C in two of them. The Frankel score improved in 10 patients, remained stable in 40 patients and worse- ned only in two patients. Preoperatively, the VAS score in 29 patients treated with PPSF procedure along with spinal decompression was 7 on average, while postoperatively, the VAS score became 5 on average. In 23 patients who underwent only PPSF procedure without spinal decompression VAS score was 5 on average, post- operatively the VAS score became 3 on average.
Xiaozhou et al 2020	Retrospective review	3	52	56.7 ± 7.4 yrs	All patients undergoing separation surgery -similar to the classic posterior pedicle approach. Following this patients had stereotac- tic body radiotherapy (SBRT). This being performed 10 - 20 days after surgery, and the target area design is generally 2–3 mm outside the tumor involvement range.	Lung 19, Breast 8, Renal 5, Thyroid 4, Liver 3, Prostate 2, Rectal 1, Gastric 1, Nasopharyngeal 1, foot rhabdomyosarcoma 1 Unknown origin 7		Three cases of postoperative complicat- ions, including two cases of wound infec- tion which improved after anti-infection and local debridement, and one case of CSF leakage which imp- roved after intraoperatively repairing the injured dura and postoperative conserv- ative treatment for 2 weeks.	FRANKEL, Karnofsky, VAS, ESCC grading and muscle grading	Postoperative Frankel neurological function grading: one case of grade A, three cases of grade B, 22 cases of grade C, 21 cases of grade D, and five cases of grade E. This compares with preoperative Frankel grades 4 cases grade A, 18 cases grade B, 24 cases grade C, six cases grade D, zero cases of grade E. Postoperative ESCC grading: 38 cases of grade 0, eight cases of grade 1a, two cases of grade 1b, three cases of grade 1c, and one case of grade 2. Preoperative ESCC grading showed one case of grade 1a, three cases of grade 1b, six cases of grade 1c, 24 cases of grade 2, and 18 cases of grade 3. The average VAS score decreased to 2.17 ± 0.52 points, which was significantly improved compared with preoperative (P < 0.01). Karnofsky performance scores: six cases of 80–100 points, 38 cases of 50 - 70 points, and eight cases of less than 50 points, which was also significantly improved compared with preoperative scores (P < 0.01).
Hamad et al. 2017	Retrospective review	3	51	60	Percutaneous pedicle screw fixation (PPSF) with or without mini- decomopression	Breast 11 (23%), Renal 9 (19%), Prostate 8 (16%), Haematological 6 (12%), Lung 4 (8%), Upper GI 3 (6%), Thyroid 2 (4%), Other 6 (12%)		One patient had septic loosening at 3 months following surgery, this required removal of implant, debridement and washout. Two patients suffered aseptic loosening at 4 months and 9 months following surgery. Implant removal alone was needed in one case. revision with open fixation in the other. One patient had unexpected major haem- orrhage (2000 mls) during mini-open decompression for metastasis from a hepatocellular carcinoma. Attempted debulking of the involved paravertebral muscles resulted in severe bleeding, which could only be controlled by clos- ure related tamponade with packs, resuscitation and return to theatre after 36 h for pack removal and re-closure. He recovered well, with improvement in KPS and pain. In the 48 remaining pat- ients the mean intraoperative blood loss was 157 mls. In patients with only a percutaneous stabilisation the mean blood loss was 92 mls, and in those with an additional decompression it was 222 mls. Overall surgical complication rate of 10%.	FRANKEL, KPS	At the time of final censoring of data, 25 patients had died (median survival of these 25 patients was 3.6 months (95% CI: 2.1–5.0). Mean follow up of patients who were still alive was 11.5 months (2–60 months). 14 patients presented with neurological symptoms, 1 with radiculopathy and 13 with incomplete cord lesions (2 Frankel B, 8 Frankel C and 1 Frankel D). Foll- owing mini-decompression, six of these 13 patients improved by one Frankel grade and none worsened. One patient had a post-operative transient foot drop that resolved spontaneously within 4 weeks. 27 patients (55%) had improvements in KPS by at least 10 points (p < 0.0005) and another 20 patients (41%) maintained their KPS score. Only 2 patients (4%) had a worsening of functional outcome, both unrelat- ed to MISS surgery. There was no significant difference in functional outcome in patients who had a fixation with decompression compared to those who had a stand-alone fixation (p = 0.33). Significant pain improvement was reported by 42 of 44 patients that presented with back pain (95%). Of the 2 patients with no improvement in pain, one pat- ient needed a proximal femoral endoprosthesis for significant post-op pain due to trochanteric metast- asis. The other patient with no change in pain had a Frankel D incomplete cord lesion. There was no diff- erence in pain outcomes in patients who had a fixat- ion with decompression compared to those who had a stand-alone fixation (p = 0.94).
Chen et al. 2000	Retrospective review	3	60	54	All patients underwent palliative surgery by ant- erior corpectomy and Zielke instrumentation. Anterior vertebral corp- ectomy allowing debulking of the occupying tumour, and reconstruction with methylmethacrylate bone cement plus Zielke's VDS internal fixator.	Lung 12 (20%), colon 10 (16.7%), liver 9 (15%), thyroid 7 (11.7%), breast 3 (5%), 1 (1.7%) for each of the following stomach, renal, nasopharynx, long bone, skin and cervical, unknown origin 13 (21.7%)		3 patients developed wound infection with localised dehiscence. In each case this was managed by local wound care. No patients suffered neurologic deterioration postop- eratively		Of the 60 patients, 4 died within 1 month, 1 each due to hepatic failure and sepsis, and 2 to respiratory failure. Five patients died within 3 months. 51 patients (85%) survived longer than 3 months postoperatively. 39 patients (67%) survived longer than 6 months, of whom 14 patients were still alive at an average of 21 months (range 13–85 months) after surgery. The average postoperative survival time for the 12 patients with lung cancer, 10 with colorectal cancer, 9 with hepatoma, and 7 with thyroid cancer documented in this series were 7, 10, 11, and 18 months, respect- ively. Neurologic improvement of at least one Frankel grade was noted in 33 patients (72%). No improve- ment was noted in the patients with a Frankel A neurologic deficit. Of the 40 patients who were bed- ridden before surgery due to pain or paresis, 30 patients experienced an increase in activity tolerance. 20 patients were able to recover functional ambulation. Sphincter dysfunction was significantly improved in 10 cases.

Table 3.

Description of surgical techniques.

Authors & Year	Classification	Evidence Level	No. of Patients	Age (yrs) ∗	Surgery Type	Primary Tumor Site		Complications	Outcome Scale	Outcomes
Zhu et al. 2020	Retrospective review	3	154	TOS group 54.10 MISS group 53.84	Separation surgery performed in a postero- lateral approach MISS under C-arm guidance	Breast (19%), Lung (18%) Kidney (10%), Liver (9%) Throid (4.5%) Colrectal (3%), Unknown (8%) Prostate (4%), Nasopharynx (5%), Other (15%)	Myeloma (4.5%)	Dural tear in one of 49 patients in the MISS group (2.04%) and two of 105 patients (1.90%) in the TOS group. Only one patient (2.04%) in the MISS group had a surgical wound infection, whereas six patients (5.71%) in the TOS group got infected.	FRANKEL	Pre-operative Frankel grades of patients in the two groups exhibited no significant difference (P = 0,59). Twenty three of 49 (47%) patients in MISS group and 43 of 105 (41%) patients in TOS group obtained improvement in neurological function (P = 0.48). 49% (24/49) and 55.3% (58/105) of patients had stable functional status after surgery in the MISS and TOS group, respectively (P = 0.47). Only two (4%) patients in MISS group and four (3.7%) in TOS group experienced worse postoperative neurological status.
Colangeli et al. 2020	Retrospective review	3	52	59.3	percutaneous pedicle screw fixations (PPSF)coup- led with mini-decompressi ions if required by spinal cord compressions.	Breast (27%), Renal (17%), Lung (8%), Prostate (2%), Other (44%)		3 minor and 3 major complications were reported. In one case an aseptic screw mobilization was detected without clinical consequences, while two patients had wound dehiscence healed without additional surgeries. As to the major complications, in two cases an intra- operative lesion of the dural sac during mini-decompression procedure was reported without consequences, in another patient a hematoma was reported with neurological compression symptoms, which required further surgical decompression resulting in symptom regression.	FRANKEL and VAS	Frankel were E in 37 patients, D3 in in 6 patients, D2 in 3 cases, D1 in 3 cases, B in one patient, and C in two of them. The Frankel score improved in 10 patients, remained stable in 40 patients and worse- ned only in two patients. Preoperatively, the VAS score in 29 patients treated with PPSF procedure along with spinal decompression was 7 on average, while postoperatively, the VAS score became 5 on average. In 23 patients who underwent only PPSF procedure without spinal decompression VAS score was 5 on average, post- operatively the VAS score became 3 on average.
Xiaozhou et al 2020	Retrospective review	3	52	56.7 ± 7.4 yrs 60	All patients undergoing separation surgery -similar to the classic posterior pedicle approach. Following this patients had stereotac- tic body radiotherapy (SBRT). This being performed 10 - 20 days after surgery, and the target area design is generally 2–3 mm outside the tumor involvement range.	Lung 19, Breast 8, Renal 5, Thyroid 4, Liver 3, Prostate 2, Rectal 1, Gastric 1, Nasopharyngeal 1, foot rhabdomyosarcoma 1 Unknown origin 7		Three cases of postoperative complicat- ions, including two cases of wound infec- tion which improved after anti-infection and local debridement, and one case of CSF leakage which imp- roved after intraoperatively repairing the injured dura and postoperative conserv- ative treatment for 2 weeks.	FRANKEL, Karnofsky, VAS, ESCC grading and muscle grading	Postoperative Frankel neurological function grading: one case of grade A, three cases of grade B, 22 cases of grade C, 21 cases of grade D, and five cases of grade E. This compares with preoperative Frankel grades 4 cases grade A, 18 cases grade B, 24 cases grade C, six cases grade D, zero cases of grade E. Postoperative ESCC grading: 38 cases of grade 0, eight cases of grade 1a, two cases of grade 1b, three cases of grade 1c, and one case of grade 2. Preoperative ESCC grading showed one case of grade 1a, three cases of grade 1b, six cases of grade 1c, 24 cases of grade 2, and 18 cases of grade 3. The average VAS score decreased to 2.17 ± 0.52 points, which was significantly improved compared with preoperative (P < 0.01). Karnofsky performance scores: six cases of 80–100 points, 38 cases of 50 - 70 points, and eight cases of less than 50 points, which was also significantly improved compared with preoperative scores (P < 0.01).
Hamad et al. 2017	Retrospective review	3	51		Percutaneous pedicle screw fixation (PPSF) with or without mini- decomopression	Breast 11 (23%), Renal 9 (19%), Prostate 8 (16%), Haematological 6 (12%), Lung 4 (8%), Upper GI 3 (6%), Thyroid 2 (4%), Other 6 (12%)		One patient had septic loosening at 3 months following surgery, this required removal of implant, debridement and washout. Two patients suffered aseptic loosening at 4 months and 9 months following surgery. Implant removal alone was needed in one case. revision with open fixation in the other. One patient had unexpected major haem- orrhage (2000 mls) during mini-open decompression for metastasis from a hepatocellular carcinoma. Attempted debulking of the involved paravertebral muscles resulted in severe bleeding, which could only be controlled by clos- ure related tamponade with packs, resuscitation and return to theatre after 36 h for pack removal and re-closure. He recovered well, with improvement in KPS and pain. In the 48 remaining pat- ients the mean intraoperative blood loss was 157 mls. In patients with only a percutaneous stabilisation the mean blood loss was 92 mls, and in those with an additional decompression it was 222 mls. Overall surgical complication rate of 10%.	FRANKEL, KPS	At the time of final censoring of data, 25 patients had died (median survival of these 25 patients was 3.6 months (95% CI: 2.1–5.0). Mean follow up of patients who were still alive was 11.5 months (2–60 months). 14 patients presented with neurological symptoms, 1 with radiculopathy and 13 with incomplete cord lesions (2 Frankel B, 8 Frankel C and 1 Frankel D). Foll- owing mini-decompression, six of these 13 patients improved by one Frankel grade and none worsened. One patient had a post-operative transient foot drop that resolved spontaneously within 4 weeks. 27 patients (55%) had improvements in KPS by at least 10 points (p < 0.0005) and another 20 patients (41%) maintained their KPS score. Only 2 patients (4%) had a worsening of functional outcome, both unrelat- ed to MISS surgery. There was no significant difference in functional outcome in patients who had a fixation with decompression compared to those who had a stand-alone fixation (p = 0.33). Significant pain improvement was reported by 42 of 44 patients that presented with back pain (95%). Of the 2 patients with no improvement in pain, one pat- ient needed a proximal femoral endoprosthesis for significant post-op pain due to trochanteric metast- asis. The other patient with no change in pain had a Frankel D incomplete cord lesion. There was no diff- erence in pain outcomes in patients who had a fixat- ion with decompression compared to those who had a stand-alone fixation (p = 0.94).
Chen et al. 2000	Retrospective review	3	60	54	All patients underwent palliative surgery by ant- erior corpectomy and Zielke instrumentation. Anterior vertebral corp- ectomy allowing debulking of the occupying tumour, and reconstruction with methylmethacrylate bone cement plus Zielke's VDS internal fixator.	Lung 12 (20%), colon 10 (16.7%), liver 9 (15%), thyroid 7 (11.7%), breast 3 (5%), 1 (1.7%) for each of the following stomach, renal, nasopharynx, long bone, skin and cervical, unknown origin 13 (21.7%)		3 patients developed wound infection with localised dehiscence. In each case this was managed by local wound care. No patients suffered neurologic deterioration postop- eratively		Of the 60 patients, 4 died within 1 month, 1 each due to hepatic failure and sepsis, and 2 to respiratory failure. Five patients died within 3 months. 51 patients (85%) survived longer than 3 months postoperatively. 39 patients (67%) survived longer than 6 months, of whom 14 patients were still alive at an average of 21 months (range 13–85 months) after surgery. The average postoperative survival time for the 12 patients with lung cancer, 10 with colorectal cancer, 9 with hepatoma, and 7 with thyroid cancer documented in this series were 7, 10, 11, and 18 months, respect- ively. Neurologic improvement of at least one Frankel grade was noted in 33 patients (72%). No improve- ment was noted in the patients with a Frankel A neurologic deficit. Of the 40 patients who were bed- ridden before surgery due to pain or paresis, 30 patients experienced an increase in activity tolerance. 20 patients were able to recover functional ambulation. Sphincter dysfunction was significantly improved in 10 cases.

Table 4.

Neurological function.

Authors & Year	Classification	Evidence Level	No. of Patients	Age (yrs) ∗	Surgery Type	Primary Tumor Site(s)	Complications		Functional Scale	Outcomes
Depreitere et al., 2020	Prospective data collection (Database)	3	914	60.7	Decompressive debulking surgery with instrumentat- ion.	Breast (18.3%), Lung (14.6%), Visceral metastases (71.8%), Extraspinal bone metast- ases (53.6%)		Post-op: Medical (3.5%), Neurological deterioration (1.9%), Wound complicat- ions (4.6%), Implant failure (0.8%), Other (7.0%), 30 day mortality (3.8%), Revision surgery within 30 days (2.9%) Intra-op: Vascular (1.9%), Neurological (1.3%), Visceral (0.4%)	FRANKEL	In 106 out of 423 patients, Frankel scores were improved at initial follow-up. (25.0%). At 6 months, patients with Frankel A - C were reduced to less than 1%, partially explained by patients fur- ther improving (14.9% between 3 and 6 months) and by a proportion of patients dying. At 12 months, however, an increase in patients with Frankel A - C was observed, due to a proportion of patients (18.8%) that deteriorated between 6 and 12 months.
Gallazzi et al. 2019	Retrospective case series	3	30	60.6 ± 11.56 yrs	Posterior-only cervical reconstruction	Lung (n = 11, 36.7%), breast and metastatic sarcomas in 4 cases each (13.3%), kidney and periph- eral nerve sheath in 2 cases (6.7%), and bladder, colon, salivary gland, skin (squam- ocellular carcinoma), stomach, thyroid, and uterus in 1 case each (3.3%).		One patient needed tracheostomy for respiratory complications after intubat- ion. One patient developed a massive pulmonary embolism and died 4 days after the intervention. During the post- operative course, 2 patients had surgic- al wound dehiscence. They both under- went revision surgery, after which healing of the wound was obtained. No sign of deep infection was observed; No instrumentation breakage was obser- ved, however, 1 patient (3.3%) had screws loostening and required revision surgery with screw placement.	FRANKEL	Only 2 patients (6.7%) had a significant worsening of neurologic function after surgery, one with a paresis of the left upper limb and one with a worsening of an already-compromised neurologic function of the upper limbs. Of the 9 patients presenting with a compromised neurologic function, 1 (11.1%) improved his neurologic status after surgery, going from Frankel grade C, to Frankel grade D. At the latest follow-up 15 patients (50%) died. All deaths except a perioperative death were due to systemic disease progression, without deaths due to local progression of the disease.
Lida et al. 2018	Retrospective	3	54	61 yrs (radia- tion group) 64 yrs (surg- ery group)	Laminectomy of the compressed spinal cord and posterior instrument stabilisation in all cases. 34 patients in the surgery group i.e surgery with or without radiotherapy. 20 patients originally asssigned to the radio- therapy alone group (radiation group)	The primary tumor histolog- ies were breast (n = 2), lung (n = 3), prostate (n = 4), liver (n = 2), kidney (n = 3), gastro- intestinal (n = 2), other gen- itourinary (n = 1), sarcoma (n = 2), and others (n = 1) in the radiation group. In the surgery group, the primary tumor histologies were breast (n = 4), lung (n = 4), prostate (n = 4), thyroid (n = 3), liver (n = 1), kidney (n = 1), pancreas (n = 2), gastrointestinal (n = 5), sarcoma (n = 5), unknown (n = 2), and others (n = −3)		There were 9 complications in the surg- ery group, i.e. 9 out of 34 (26%). Post- erative paraplegia in one, hematoma in two, wound infection in three, post- operative pneumonia in two, and gastric perforation in one patient. Death from rupture of liver metastasis occurred in one patient.	FRANKEL	Seven patients initially in the radiation group underwent surgery because of a substantial decline in their motor strength during radiotherapy. One of the remaining 13 patients (8%) in the radiat- ion group and 30 of the 34 patients (88%) in the surgery group showed improve- ment in their neurological symptoms (P < 0.01). One patient (8%) in the radia- tion group and 21 patients (62%) in the surgery group were ambulatory after treatment (P < 0.01). Four of the seven patients converted to surgery in the radiation group showed improvement in their neurological symptoms. One of 8 (13%) patients in the radiation group and 6 of 16 patients (38%) in the surg- ery group who were initially not ambul- atory regained the ability to walk after treatment (P = 0.352). The 50% survival rate was 113 days in the radiation group and 365 days in the surgery group (P = 0.03). The Frankel grade was A in 1, C in 8, and D in 11 (better than D: 55%) in the radiation group, and C in 16 and D in 18 (better than D:52%) in the surgery group.
Pessina et al. 2018	Retrospective	3	97	59	Palliative decompression was performed in 27% of patients, tumor curett- age (debulking) was perf- ormed in 51%, and total vertebrectomy was perf- ormed in 22%, followed by radiotherapy in 78% of cases.	breast (n = 23), lung (n = 30) GI (n = 12), GU (n = 23), other (n = 7)		Minor or major perioperative morbidity occurred in 14 (14.4%) patients: 7 had perioperative complications (CSF leak- age, thoracic duct injury, and dysphagia), 2 had implant failure, and 5 had wound complications. A surgical perioperative mortality occurred in 7 patients. Medical complications, such as pneumonia, deep vein thrombosis, and urinary tract infect- ion occurred in 7.6% of patients.	FRANKEL	Nearly all patients (95.9%) had back pain before treatment, with a VAS score of 7 or greater in 81.7% of cases. Major and minor preoperative neurologic deficits (grades A - D on the FS) were present in 73 cases (75.3%). No patients had complete loss of both motor and sensory function (grade A on the FS).Complete loss of motor function with some preservation of sensory function was found in 8 cases (8.3%) (grade B on the FS), 33 (34%) had lost all sensory function with some motor function preserved though not enough to be functional (grade C on the FS), 32 (33%) had weak but useful motor function (grade D on the FS), and 24 (24.7%) were neurologically intact (grade E on the FS). Postoperative neurological function improved in 46 patients (51.1%), and it remained stable in 44 patients (48.9%). Among patients who had normal motor function preoper- atively (grade E on the FS), all retained this function postoperatively.
Li et al., 2018	Retrospective	3	34	54	40 procedure were perform- ed in 34 patients. 8 patients underwent anterior approach alone (the median blood loss was 140 ml), 22 patients underwent post- erior approach alone (the median blood loss was 320 ml, 4 patients underwent combined posterior - anterior approach (the med- ian blood loss was 580 ml).	Lung 13 (38.2%), Breast 7 (21.9%), Prostate 6 (17.6 %), Gastrointestinal 2 (5.9 %), Other 6 (17.6%)		4 complications (11.8%), with 2 (5.9%) requiring reoperation. 1 patient (2.9%) experienced acute epidural hematoma after posterior surgery which was treated by surgical treatment with no neurolog- ical deterioration. 1 patient (2.9%) exp- erienced deep wound infection after posterior surgery which was success- fully treated by surgical debridement. 1 patient (2.9%) had cardiorespiratory worsening following anterior or combined posterior-anterior surgery. One patient (2.9%) had postoperative cerebrospinal fluid leakage/effusion confirmed by MRI, whereas 1 year follow-up did not show any special symptom. There were no instrumentation failures in this study.	FRANKEL	Majority of patients (71%) maintained or improved their Frankel scores 1 year after surgery. KPS scores improved in 13 patients (38%), remained stable in 19 (56%), and worsened in 2 (6%) postop- eratively. Notably, patients with neuro- logical deficit that did not improve after surgery had significantly worse median survival than those who had either no deficit or who improved after surgery. The entire patients’ median survival time was 12.4 months (range 3.5–36.2 months). Pain improved in 32 patients (94.12%), and the postoperative VAS scores were significantly improved compared with preoperative data.
Uei et al. 2018	Retrospective	3	55	66.8	posterior only instrument- ation.	lung 11 (20.0%), liver cancer 9 (16.3%), prostate cancer 6 (10.9%), myeloma 5 (9.0%), kidney 4 (7.2%), thyroid 4 (7.2%), lymphoma 3 (5.4%), gallbladder 3 (5.4%), breast 2 (3.6%), sarcoma 2 (3.6%), others 5 (9.0%), unknown 1 (1.8 %)		Perioperative complications included massive bleeding of ≥1500 ml during surgery in 5 patients (9.0%), death with- in 30 days after surgery in 3 patients (5.4%), epidural hematoma after surgery in 2 (3.6%), and wound dehisce- nce in 2 (3.6%). With regard to two post- operative epidural hematoma patients, they were treated with surgical evacu- ation immediately after diagnosis. As for two patients with wound dehiscence, they were treated with resuture.	FRANKEL	The grade of preoperative paralysis on the modified Frankel scale was A in 2 patients (3.6%), B in 1 (1.8%), C in 19 (34.5%), D1 in 15 (27.2%), D2 in 1 (1.8%), D3 in 10 (18.1%), and E in 7 (12.7%). Regarding improvement of paralysis, the modified Frankel scale was improved by one grade or more or grade E was maintained in 35 patients (63.6%), whereas paralysis aggravated in 2 (3.6%). The median postoperative survival time determined using the Kaplan-Meier method was 12.0 months (95% confid- ence interval 2.4–21.5).
Cofano et al. 2020	Retrospective	3	84	66.5	Comparison of outcomes associated with Anterior/ anterior-lateral (AD); posterior/posterior-lateral (PD/PDL); circumferential (CD) decompressions.	NSCLC 19 (22.6%), myel- oma 16 (19.0%), breast 9 (10.7%), colon 6 (7.1%), prostate 8 (9.5%), lymphoma 9 (10.7%), renal 5 (6.0%), melanoma 2 (2.4%), thyroid 3 (3.6%), SCLC 2 (2.4%), stomach 2 (2.4%), others 3 (3.6%)		It is tabulated that there were a total of 10 (11.9%) complications. This has not been elaborated on, in the article.	ASIA	AD/CD patients showed higher chance of neurological improvement and red- uced rates of worsening compared to PD/PLD group (94.1%/100% vs 60.4%; 11.8% vs 45.8% respectively). Univariate logistic regression identified immediate postoperative improvement to be a sig- nificant protective factor for worsening at last follow-up. Stratifying patients for site of compression and considering anterior and circumferential groups, immediate post-operative neurological improvement, was mostly associated with AD and CD (p 0.011 and 0.025 respectively). Walking at last follow up was influenced by post-operative maint- enance of ambulation (p 0.001). According to neurologic assessment, 84 patients reported a pre-operative neur- ologic deficit. Immediate post-operative evaluation reported neurologic improve- ment in 64 patients (76.2%), stability in 19 patients (22.6%), while in 1 case (1.2%) a worsening was observed. At follow-up (Mean 10.8 months) a deter- ioration of neurologic status was recorded in 28.6% of all the cases (24 patients). The most common timing of deterioration was registered within 3–12 months after surgery (3–6, 20.4%; 6 - 13, 18.3%).
Lak et al., 2020	Retrospective	3	151	60.4	Posterior decompression (90.5%), Anterior approach (7.43%), Combined anter- ior-posterior approach (2.0%).	Lung (18.9%), multiple myeloma (15.5%), renal cell cancer (13.5%), breast cancer (12.8%), pros- tate cancer (10.1%), Gastrointestinal cancers (8.10%), lymphoma 3.37%), gynecological tumors (2.0%), melanoma (2.0%), Others (<15%),		Except for 3 complications (2 wound in- fections/dehiscence at the 2-month follow up and 1 hardware failure at the 3 year follow-up), all complications ocurred in the 30-day postoperative period. The most common complications were wound infection/wound dehiscence, systemic infections, dural tears/CSF leaks, fluid/electrolyte disturbance, deep venous thrombosis(DVT)/pulmonary embolism (PE), excessive blood loss, instrument complic- ations (including hardware failure), and post-operative hematoma causing cord compression (1 patient)	ASIA and Eastern Cooperative Oncology Group (ECOG) performance status grades	With the ASIA scale, most patients were classified as grade D (61.2%) preoperat- ively, followed by grade C (23.4%), grade B (11.0%), and grade A (4.13%). Postop- eratively, ASIA grade E (53.2%) was the most common, followed by grade D (24.0%), grade C (12.4%), grade B (6.56%), and grade A (3.64%). In particular, there were 6 patients with preoperative ASIA grade A impairment. Of these 6 patients, 1 patient improved to grade B, and another to grade D, the rest remained at grade A. Similarly, of the 16 patients who presented with ASIA grade B, 7 improved to grade C, 1 to grade D, 2 to grade E, 4 remained as grade B, and 2 patients had no data postoperatively. Regarding the 34 patients who presented with ASIA grade C, 9 improved to grade D and grade E, 7 remained as grade C, 3 progressed to grade B, 1 progressed to grade A, and 5 patients had no data postoperatively. For patients presenting with ASIA grade D impairment (n = 89), 59 improved to grade E, 22 remained as grade D, 3 pat- ients progressed to grade C, 1 patient progressed to grade B and 4 patients had no postoperative data. The mean pre- and postoperative ECOG functional status grades were 3.2 ± 0.8 and 2.4 ± 1.1 respectively

Table 5.

Miscellaneous outcomes.

Authors & Year	Classification	Evidence Level	No. of Patients	Age (yrs) ∗	Surgery Type	Tumor Histology	Complications (no.) or %	Outcomes
Chong-chung et al., 2018	Retrospective study	3	65	57.26	A total of 29 (44.62%) patients had the Luque rods and sublaminar wire system, 24 (36.92%) pat- ients had posterior screws and rods system, 9 (13.85 %) patients had an anterior cage and plate applied, and 3 (4.62%) had the combined anterior & posterior system.	17 (26.15%) patients suff- ered from lung primary, while 16 (24.62%) patients suffered from breast prim- ary. Other malignant origins included colorectal, thyroid, prostate, renal cell carcin- oma, nasopharyngeal car- cinoma, hepatocellular carcinoma, germ cell tum- our, multiple myeloma, bladder carcinoma, hypo- pharynx carcinoma and soft tissue sarcoma.	This paper focused on loostening rate and consequences of this. The overall loosening rate of imp- lants was 44.4%. This was particularly high in the group with Luque rods and sublaminar wire system (70%) and the posterior screws and rod system (50%).	Survival ranged from 1 month to 92 months, with a mean survival of 13.5 months. 40 (61.54%) of the 65 cases survived for at least 1 year after the operation. 27 (67.5%) of them had imaging after 1 year postoperatively, mean survival here of 27.63 months; the remaining 13 patients had no available imaging after 1 year post- operatively, here the mean survival was 23 months.
Walter et al. 2012	Retrospective study	3	57	58.6	All patient were operated in the prone position. For posterolateral decomp- ression, laminectomy is performed and bilateral removal of the pedicles and facets to achieve circumferential decomp- ression and facilitate post- erolateral tumour excision. To improve seeing the thoracic spine, the medial rib head is excised, if needed. After adequate decompression, the operated vertebrae are stabilized, using the UniFlex screw rod system.	Plasmocytoma 10 (17.6%), breast 8 (14.0%), lung 7 (12.4%), renal 6 (10.6%), prostate 5 (8.8%), CUP- syndrome 4 (7.0%), colon 3 (5.3%), cervix 3 (5.3%), bladder 2 (3.5%), NHL 2 (3.5%) and 1 (1.7%) each for chordoma, ovary, pancreas, sarcoma, stomach, testis, undiffer- entiated carcinoma.	The complication rate was only 5.3% with 2 superficial wound infections along with 1 seroma likely secondary to previously received radiotherapy. One patient required repeat operation for second-level decompression and stab- ilization because of a massive local tumor recurrence. No intraoperative death occurred, but 3 (5.3%) patients died within 4 weeks of surgery, secondary to rapid progression of the primary metastatic disease. No case of posterior spinal instrumentation fatigue failure was detected.	The mean postoperative survival of all 57 patients was 11.4 months and ranged from 0 to 46 months. The 1-year survival rate was 42.1%. There being a tendency toward improved Frankel grades after the operation. 13 (22.8%) patients showed an improvement of their neurological deficits, 43 (75.5%) had a stable neurological status, and only one single patient (1.8%) suffered from deterioration of neurological status.

3.2. Predictors of survival

Eight studies provided results for predictors of survival in patients with spinal metastases, who subsequently went on to have some form of decompression surgery (Table 1),.^{1,2,3,4,5,6,7,8} Of these, 6 studies were retrospective, 1 prospective and 1 was a semi-prospective study (containing partly retrospective and partly prospective data). Surgical interventions in this category included decompression alone, decompression with instrumentation, from an anterior, posterior or combined approach and surgery with adjuvant radiotherapy. In these included studies, the primary histology of tumors showed considerable variability. Most commonly described were lung (7 studies), breast cancer (6 studies), prostate cancer (6 studies), and renal (6 studies).

In a semi-prospective study comprising 201 patients by Quraishi et al. where the authors are evaluating the usefulness of the Tokuhashi scoring system in predicting prognosis and decision making following surgery for metastatic spinal cord compression,⁸ patients were divided in to three groups according to their Tokuhashi score - Group 1 (Tokuhashi score 0–8, n = 84), Group 2 (Tokuhashi score 9–11, n = 83) and Group 3 (Tokuhashi score 12–15, n = 34). This identified no significant difference in the neurological status (Frankel grade) between Groups 1 and 2 (p = 0.34) or Groups 2 and 3 (p = 0.70). However, there was a significant difference between Groups 1 and 3 (p = 0.001), clearly Group 3 showed their neurological outcome to be much better. Median survival was 93 days in Group 1, 229 days in Group 2 and 875 days in Group 3 (p = 0.001). The predictive value between the actual and predicted survival was 64% (Group 1), 64% (Group 2) and 69% (Group 3). The overall predictive value of the revised Tokuhashi score using Cox regression for all groups was 66%.

Kobayashi et al.² reviewed the merits of the Katagiri score in helping to predict survival. Of the 201 patients they reported on, 29 had posterior decompression alone and 182 patients had instrumented posterior decompression. On the basis of survival analysis, they used information about primary tumour site, visceral metastasis presence, laboratory data, performance status and chemotherapy information to obtain this new score. In univariate analysis, moderate growth and rapid growth at the primary site were significantly associated with poor survival, with HRs of 3.11 (95% CI, 2.65–5.29, P < 0.01) and 4.97 (95% CI, 3.38–9.53, P < 0.01) compared with reference data, respectively. Nodular and disseminated metastasis were also significantly associated with poor survival, with Hazard ratios (HRs) of 1.85 (95% CI, 1.28–4.21, P < 0.01) and 3.12 (95% CI, 1.62–5.53, P < 0.01), respectively. In multivariate analysis, the same variables continued to be associated with poor survival: moderate growth and rapid growth at the primary site, with HRs of 2.95 (95% CI, 1.27–7.89, P < 0.01) and 4.71 (95% CI, 2.78–12.31, P < 0.01), respectively; nodular and disseminated metastasis, with HRs of 1.53 (95% CI, 1.07–3.85, P < 0.01) and 2.94 (95% CI, 1.33–5.42, P < 0.01), respectively. The authors concluded that of the new Katagiri variables, growth at the primary site, visceral metastasis, critical laboratory data, and poorer performance status were all clearly significant independent prognostic factors showing a correlation with decreased survival in patients with spinal metastases.”

Jansson et al.⁸ found no significant difference when considering the surgical approach (either anterior or posterior) and predicted survival. This study like most others was heavily biased towards a posterior approach. Here, 282 patients in this series had surgery, 212 had posterior decompression and stabilisation, demonstrating a heavy bias towards this approach. In 47 patients, only a posterior decompression was performed and only 23 patients had an anterior decompression alone. The survival rate was 0.63 at 3 months, 0.47 at 6 months, 0.30 at 1 year, and 0.16 at 2 years. In this study, 5% (n = 12) of patients with motor deficits worsened postoperatively, whereas 70% (n = 179) improved at least one Frankel grade. They concluded that neurological improvement in function can be achieved by surgery. In this study, the complication rate was high (60 complications in 56 of the 282 patients) and many patients also died of their disease within the first months of surgery.

Hohenberger et al.⁴ analysed 94 patients undergoing decompressive surgery for spinal metastases in a retrospective manner. Their univariate analysis indicated that the risk of poor outcome defined as Karnofsky Performance Index less than 70 at discharge was significantly higher for male patients (p = 0.001) and for patients with multiple spinal metastasis (p < 0.001), pre-operative urinary retention, and loss of sphincter control.

Other studies also sought to determine the accuracy and reliability of scoring systems for predicting the prognosis and survival. Perhaps the most well recognised being the tokuhashi score. De Almeida et al.⁵ reviewed the reliability of the tokuhashi score in a population of 117 patients who had surgical decompression for metastatic spinal disease. Their analysis concluded that the tokuhashi score was not an accurate tool for prognosis prediction. Patients with lower scores where surgery was not indicated by the tokuhashi score had an improved quality of life and longer survival after surgical intervention. They did not find this score to provide an accurate means to establish the optimal treatment and prognostic indicator for their patients.

Conversely in a review by Vanek et al.,⁶ they retrospectively analysed 166 patients having surgery for symptomatic spinal metastases. They considered the following parameters: “age, primary tumor aggressiveness (slow, moderate, rapid growing), spinal location (cervical, thoracic, lumbar, and sacral), operation type (posterior decompression, anterior or posterior instrumented procedure, and radical combined instrumented surgery), preoperative evaluation using the revised Tokuhashi scoring system (<9, 9–11, and 12–15), pre- and postoperative neurological status according to the Frankel score (A-C and D-E), and the site of the main spinal cord compression (anterior, posterior, or combined).” Their results showed that the median survival time following surgery was 16.0 months. Preoperative neurological status influenced survival time significantly; the median survival was 5.1 months in Frankel A-C and 28.2 months in Frankel D-E (P < 0.001). Improvement on the Frankel scale had no correlation with the survival time (P = 0.131). When the patients' age was <65 years this related to a significantly longer survival time (P = 0.046). The Tokuhashi score predicted patient's survival independently (P < 0.001). Other analysed parameters did not demonstrate statistical significance. Hence in their review the salient considerations in the postoperative survival of patients with symptomatic spinal metastases comprised their preoperative neurological status and the Tokuhashi scoring system, when these two factors are combined, in their opinion this provides the clinician with a practical and useful way for planning the extent of surgical treatment.

3.3. Predictors of ambulatory status or motor function

Only eight articles directly report on factors that affect ambulatory status after surgery or motor recovery following decompression surgery for spinal metastases, Table 2,.9, 10, 11, 12, 13, 14, 15, 16 Of these, 6 were retrospective and two prospective reviews.

Rades et al.,⁹ reviewed patients who underwent surgery with radiotherapy, S + RT, (n = 67) versus radiotherapy, RT, alone (n = 134). The local control rates for the entire cohort was 93% at 6 months and 86% at 12 months. An improved motor score was found in 22% of patients after S + RT and 16% after RT alone (p = 0.25). Post-treatment ambulatory rates were 67% and 61%, respectively (p = 0.68). Of non-ambulatory patients, 29% and 19% (p = 0.53) regained ambulatory status. This study incorporated a matched-pair analysis and identified that individuals who had metastatic epidural spinal cord compression (MESCC) from more unfavourable, radioresistant primary tumors (e.g., renal cell and colorectal carcinoma) had an improved functional outcome with decompressive surgery, stabilisation and radiotherapy, this was not the case when only laminectomy and radiotherapy was performed.

In a group of 101 patients undergoing decompressive laminectomy for spinal metastases, reviewed by Younsi et al.,¹⁰ at discharge, 83 patients (82%) stated an overall improvement in their symptoms. Nerve root palsies showed a good recovery (improvement in 73% of cases) and pain relief was provided in most (radiating pain in 54% and back pain in 47% of cases). However, sensory deficits as well as bladder or bowel dysfunction were often persistent (improvement in 18%, 24%, and 20% of cases, respectively). Pre-operatively impaired neurological function (Frankel Grade A–D) had improved by ≥ 1 grade in 61% of patients at discharge. It was noted that, 25% of all severely impaired patients (Frankel Grade A and B prior to surgery) and 51% of all non-ambulatory patients (Frankel Grade A–C) had regained ambulation after surgery. Overall, 61 patients (60%) were ambulatory at discharge (Frankel Grade D and E) compared to 20 patients (20%) prior to surgery. A better post operative functional status was seen in the KPI score in 75 patients (74%). On discharge, 27% of patients the KPI score ≥80 compared to 4% prior to surgery.

The ability to ambulate as well as an assessment of the neurological status are important indicators to determine suitability for postoperative treatment including chemotherapy as highlighted by Tateiwa et al.¹¹ Here 31 patients underwent a posterior approach to allow for the direct circumferential decompression of the spinal cord with or without subsequent stabilisation. 21 patients (68%) improved by at least one Frankel grade; 17 patients (55%) became ambulatory postoperatively. Most postoperative ambulatory patients could undergo postoperative chemotherapy (14/17, 82%). In contrast, only a few postoperative non-ambulatory patients could undergo chemotherapy postoperatively (2/15, 13%). The complication rate was 35.5%, specific complications were wound infection, pneumonia, and deep vein thrombosis/pulmonary embolus. “The median survival duration was 7.0 months. Factors that significantly affected the overall survival in univariate

analyses were revised Tokuhashi score (RTS) ≥ 4, postoperative chemotherapy, ambulatory status, and complications (RTS ≥4, P < 0.05; postoperative chemotherapy, P < 0.001; ambulatory status, P < 0.001; complications, P < 0.01).” The outcomes therefore suggested that surgery for patients who are non-ambulatory due to MESCC directly correlates with functional outcomes and may have an indirect contribution to overall survival. Situations where non-ambulatory patients who are assessed as unable to tolerate chemotherapy due to their poor performance status subsequently regain the ability to walk by decompressive surgery, have a chance to receive postoperative chemotherapy, thereby increasing their chances of prolonging survival.

Itshayek et al.¹² identified a significant relation between duration of ambulation and both preoperative (p = 0.0342, Kruskal-Wallis; p = 0.0367 log-rank) and postoperative ASIA grade (p = 0.0358, Kruskal-Wallis), as well as a possible trend toward significance between preoperative ASIA grade and survival (p = 0.0886, log-rank). The importance of preoperative motor function was further highlighted by Lo et al.,¹³ Here, patients who had an intact motor status preoperatively demonstrated improved survival, (Group A, n = 37/89) compared to those with motor deficit (Group B, n = 52/89, p = 0.0031). In Group B, survival was improved when surgery was performed within 7 days of motor deficit onset as opposed to when performed 7 days after onset (p = 0.0444) and in postoperative ambulant patients than in non-ambulant patients (p = 0.0120).

Kwan et al.¹⁴ prospectively reviewed (n = 50) a minimally invasive technique of stabilisation (MISt) using fluoroscopic guided percutaneous pedicle screws with (74%) and without (26%) minimally invasive decompression. They identified that 70% showed improvement of one frankel grade, 5% improved two frankel grades. No patient was bed-ridden postoperatively, with the average time to ambulation of 3.4 ± 1.8 days. MISt showed a statistically significant reduction in visual analog scale pain score with mean preoperative score of 7.9 ± 1.4, this was significantly decreased to 2.5 ± 1.2 postoperatively (p = 0.000).

Park et al.¹⁵ also identified a strong correlation between preoperative ambulation and postoperative ambulation, namely that grade three lower extremity power was a significant criteria to achieve postoperative ambulation. Also that pre-operative lower limb power (p < 0.001) along with pre-operative ambulation (p < 0.001) are significant predictors of postoperative ambulation.

3.4. Description of surgical technique

Whilst the traditionally recognised surgical technique for decompressing metastatic spinal cord tumors appears to be a posterior approach, namely open decompression with or without stabilisation, five studies^{17,18,19,20,21} (see Table 3) describe alternative techniques or variations of the posterior approach and outcomes associated with these. All five studies were retrospective data collections. In all five, breast and lung cancers were described as primary tumors in the reviews, prostate along with renal cancer was described in 4 studies.

Zhu et al.¹⁷ divided patients (n = 154) in to those having Traditional Open Surgery (TOS) (n = 105) and those having Minimally Invasive Spine Surgery (MISS) (n = 49). Key points from this review include that the mean intraoperative blood loss in the MISS group was lower than in the TOS group (748.57 vs 950.48 ml, p = 0.039). Wound infection rate was lower in MISS group (2.04%) than that in TOS group (5.71%), however this difference was not significant. The MISS group had a significantly shorter hospital stay compared with TOS group (7.35 vs 9.94 days, p = 0.0007). Both MISS and TOS group had comparable operative duration and improvement of neurological function. Therefore, the study concluded that MISS would be an excellent alternative for the surgical management of spine metastases.

In another analysis of the use of MISS in metastatic spinal surgery, Hamad et al.²⁰ performed a review of percutaneous pedicle screw fixation (PPSF) with and without mini decompression in 60 patients. Karnofsky performance status (KPS) was the primary functional outcome, their study showed no significant difference in functional outcome when fixation with decompression was performed compared to when fixation alone was implemented (p = 0.33). In this review 13 patients had an incomplete cord lesion preoperatively (2 Frankel B, 8 Frankel C and 3 Frankel D). With mini-decompression, of these 13 patients six showed an improvement by one Frankel grade, no patient deteriorated. Postoperatively there was one case of a transient foot drop, this resolved spontaneously in the next 4 weeks.

In a recent review by Colangelli et al.¹⁸ (2020), 52 patients also underwent PPSF for metastatic spinal surgery, the authors reported that patients had an improvement of neurological status and pain with a relative low complication rate. They also stated that although the efficacy of PPSF is well recognised in trauma as well as degenerative spine surgery, there is insufficient literature in relation to MISS techniques being used in spinal metastasis Hence further evidence and research are needed with this technique when applied to metastatic spine.

Chen at al²¹ analysed 60 patients who had anterior reconstructive spinal surgery with zielke instrumentation for metastatic malignancies of the spine. They advised that Indications for this anterior approach included severe pain not responsive to strong narcotic analgesics (52 patients), neurologic deficits (46 patients) and unstable spine (36 patients). Most patients had been administered either chemotherapy or radiotherapy prior to review with a spinal specialist and had not shown any neurological change. Patients were already paraparetic or paraplegic, leading to a bedridden and a very painful back, hence the authors hoped that anterior decompression and fixation could provide pain relief, maintain spinal stability, and restore the patients’ neurologic status. Their results showed that 52 patients had symptoms of severe pain pre-operatively, 40 of them (77%) obtained pain relief for a certain period before death. 46 patients (76%) presented with neurologic deficits. Their thoracic levels tended to be more profound in nature. The study concluded that in cases of anterior body destruction with spinal cord compression secondary to tumor metastasis, anterior decompression and anterior reconstruction with bone cement and Zielke ventral derotation spondylodesis (VDS) fixation can relieve pain, improve neurologic function, and achieve spinal stability.

3.5. Neurological function

To be considered as part of the inclusion criteria for this review, studies were required to have some mention of neurological function, with at least mention of Frankel grades or ASIA scores. Table 4 identifies eight studies^{22,23,24,25,26,27,28,29} where neurological function forms the major component of the authors’ review. 7 were retrospective and 1 a prospective study. Three of these studies reported the use of an anterior approach, in some patients in combination with a posterior approach, the others report the use of a posterior approach alone.

A subgroup analysis by the global spine tumor study group by Depreitere et al.²² reviewed 914 patients who underwent decompressive debulking surgery with instrumentation who despite noting an initial improvement in post-operative frankel scores (25% at initial followup), by 12 months follow up there was an increase in patients with Frankel A - C, due to a proportion of patients (18.8%) that deteriorated between 6 and 12 months.

Gallazzi et al.²³ produced what they described as the largest series of patients treated with a posterior-only approach for cervical spine metastasis, achieving satisfactory postoperative pain control as well as spine stability. In so doing, they stated the posterior approach and fixation in cervical spine metastasis could be a safe alternative to an anterior or combined approach option, even in cases with anterior column involvement.

Lida et al.²⁴ retrospectively reviewed the neurological outcomes associated with radiotherapy (n = 20) versus surgery (n = 34) in patients with MSCC with a presentation of myelopathy. They identified that radiotherapy when given by itself was not as effective as surgery in MSCC patients in this setting. Surgery is often required to recover neurological function. One should be aware that there is a high incidence of complications that should be considered. Also noting that the symptoms and compressive level of the spinal cord should also be considered when comparing the neurological outcomes between radiotherapy and surgery, as the ideal treatment can vary from one scenario to the next.

Cofano et al.²⁸ performed a review of 84 patients to analyse whether the type of decompression performed does in fact influence the neurological outcome in patients with spinal metastases. Type of decompression was sub-divided in to “(anterior/anterior-lateral (AD); posterior/posterior-lateral (PD/PDL); circumferential (CD). A total number of 84 patients were included. AD/CD patients had an increased incidence of improved neurology along with reduced rates of deterioration compared to the PD/PLD group (94.1%/100% vs 60.4%; 11.8% vs 45.8% respectively). The requirement to eliminate the source of epidural metastatic compression should be regarded as a priority. Since the majority of spinal cord compression involves firstly the ventral part of the sac, hence for this reason CD/AD are associated with better neurological outcomes and should be targeted in situations where there is circumferential or anterior/anterolateral compression.”

Lak et al.,²⁹ aimed to quantify the results of decompressive surgery on patients quality of life in symptomatic metastatic spinal disease. They reviewed 151 patients where the majority of patients had a posterior approach for their spinal metastases (90.5%). The mean pre- and postoperative ECOG (Eastern Cooperative Oncology Group) performance status grades were 3.2 and 2.4, respectively. 58.3% of patients at follow-up demonstrated improvement, 31.5% showed no change, and 10.0% demonstrated a deterioration of functional status. The authors identified the mean QALY gained from surgical decompression in the first 6 months and first year equalled 1.2 months and 5 months respectively (of life in perfect health). These findings giving support to surgical intervention in situations where life expectancy <6 months. With the ASIA Impairment Scale, most patients were classified as grade D (61.2%) preoperatively, followed by grade C (23.4%), grade B (11.0%), and grade A (4.13%). Postoperatively, ASIA grade E (53.2%) was the most common, followed by grade D (24.0%), grade C (12.4%), grade B (6.56%), and grade A (3.64%). The authors therefore reporting that surgical decompression provides considerable chances of neurological recovery and good functional performance in patients presenting with neurological deficits from MSCC.

3.6. Miscellaneous outcomes

Two studies reviewed outcomes not directly related to the topics already described^30,31 (Table 5). ChongChunget al.³⁰ concentrated on the loosening rate of implants in patients undergoing decompression surgery for spinal metastases. They reviewed data from 65 patients. The survival at one year post-operation was 61.54%. The overall loosening rate of implants was 44.44%, Luque rods and sublaminar wire system being the most (70%) commonly affected systems. No cases needed revisional surgery or implant removal at 1 year postoperatively. The authors postulated that the loosening rate of implants was high, and would be expected to grow even higher as oncological patients continue to show improved level of; survivorship with medical advances, for example target therapy. There remains (at present) no definitive studies on how the loosening of implants would impact patients' quality of life and clinical performance.

Walter et al.³¹ reported on the outcomes of patients with metastatic spinal disease, treated with palliative considerations using the techniques of spinal decompression and posterior instrumentation. In this study 57 individuals had a posterolateral approach for decompression and posterior instrumentation. The authors described excellent clinical outcomes. Namely all patients with intractable pain showed significant improvement postoperatively, there was no neurological decline seen. As those with spinal metastatic disease approach the terminal stage of their disease, it is generally accepted that they will need only palliative surgical treatments. Hence, in these situations spinal decompression and stabilisation may be employed to enhance the quality of the remaining life of cancer patients.

3.7. Complications

Given the comprehensive nature of this review, it is not surprising that the complications reported in the studies reviewed are extremely varied. By far the most frequently reported complication was wound infection or wound dehiscence (25 studies). It was noted that in some studies either surgical complications were not stated, namely by Kobayashi et al.,² or where stated they were not expanded on, de Almeida et al.⁵ as well as Cofano et al.²⁸ who reported complications occurring in 10 patients (11.9%) however, this data is not elaborated on. Jansson et al.⁹ reported an increased intraoperative time along with a raised blood loss increased the rate of postoperative infection. Hence, when surgery was performed for more than 4 h, this was associated with a threefold increased rate of infection and a blood loss of greater than 3000 ml had a twofold increase. In this article wrong level surgery was also reported, namely two patients with epidural compression at T3 and T4, respectively, had wrong level decompressive surgery performed, hence in both cases they were re-operated the following day. One remained paraplegic, the other regained normal motor function and survived 6 years. Where documented in these studies, systemic/medical complications tended to concentrate on cardiorespiratory events particularly deep vein thrombosis, pulmonary embolism and post-operative pneumonia.

4. Conclusion

This work represents a comprehensive systematic review of outcomes following decompressive surgery for metastatic spinal tumors. In particular we highlight those articles that emphasise salient predictors of survival, ambulatory status along with motor function. Survival data and neurological function (pre and post-operatively) is relayed in a succinct fashion for the studies presented. Where appropriate we also provide a review of the surgical techniques used to achieve decompression with and without stabilisation for patients with metastatic spinal cord compression.

CRediT authorship contribution statement

Bhoresh Dhamija: Conceptualization, Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Dheeraj Batheja: Data curation, Formal analysis. Birender Singh Balain: Conceptualization, Methodology, Supervision.

Declaration of competing interest

The authors report no conflicts of interest.