Safety and risk analysis of total resection surgery for thoracic and lumbar spinal tumors: a decadal analysis of 103 cases

Abstract

Study design

Retrospective cohort study.

Purpose

To explore the complications and risk factors for total resection (TR) of primary thoracic and lumbar spinal tumors over the past decade at our institution.

Methods

Patients meeting inclusion criteria (primary spinal tumors, thoracic or lumbar location, TR at our center) were included. Demographic characteristics, surgical data, perioperative complications and management results were reviewed. Patients were stratified by tumor site, the number of excised segments, and recurrence status to elucidate distinctive characteristics.

Results

The cohort comprised 103 patients, with a mean age of 35.8 years. On average, 1.83 vertebral segments were resected per patient. Perioperative complications were substantial, totaling 166 events, with 71 classified as major and 95 as minor, yielding an average of 1.61 complications per patient. No perioperative deaths occurred, but 79 patients (76.7%) experienced at least one complication. Multiple vertebral sections correlated with a higher complication rate (P = 0.031), and lumbar surgeries exhibited elevated risks of large vascular injury (P = 0.001), neurological deterioration, and cerebrospinal fluid leakage compared to thoracic cases. Conversely, thoracic spinal procedures showed a higher rate of pleural effusion (P = 0.004). Binary logistics stepwise regression identified multi-segmental resection as the independent risk factor for major perioperative complications.

Conclusions

TR of primary spinal tumors is associated with a high perioperative complication rate, although most events have a favorable prognosis. Complication characteristics vary based on the surgical site, number of excised segments, and surgical history. A nuanced preoperative evaluating approach considering patient age, surgical segments, and extent of resection is crucial.

Introduction

The management of primary spinal tumors poses a unique set of challenges due to its rarity, constituting 2.8%-13% of bone tumors. With an annual occurrence rate of 2.5–8.5 per 100,000 people [1], these tumors demand specialized attention. The primary treatment approach for most spinal tumors involves the complete removal of lesions, typically executed with a tumor-free margin through wide or marginal radical surgery. Tomita et al. pioneered total en-bloc spondylectomy (TES) techniques, utilizing a T-saw to achieve tumor-free margins within the vertebral body [2, 3]. TES has become a significant surgical option, offering the potential for complete tumor removal in carefully selected patients.

While TES aims to achieve wide or marginal R0 resection, it is associated with both significant advantages and notable challenges. The procedure allows for the precise removal of the tumor with minimal disturbance to surrounding tissues, thus reducing the risk of local recurrence. However, TES is technically demanding and comes with substantial risks, particularly when tumors involve the pedicles or extend into surrounding critical structures. Boriani et al. further refined surgical approaches by categorizing seven subgroups for total en-bloc resection (TER) of the tumors, broadening the applicability of TES in diverse anatomical scenarios [4, 5]. Despite these advancements, surgical outcomes remain variable, with complication rates reported between 46.2% and 86.7% for spinal tumor resections [6–11]. Given the spine's proximity to vital structures like the spinal cord, nerves, pleura, peritoneum, abdominal organs, and large vessels, complications can arise, including wound healing issues, cerebrospinal fluid leakage, neurological deterioration, and even major vascular injuries.

Despite the advances in TES and TER, several challenges remain, particularly in achieving optimal surgical outcomes in complex cases where tumor location impedes total resection (TR). When tumors invade cervical, cervicothoracic, or lower lumbar spine regions, achieving wide or marginal R0 resection becomes exceedingly difficult. Additionally, in patients with complex anatomy, poor general condition, or where technical limitations exist, intralesional tumor removal may be the only viable option. These unresolved challenges underscore the need for further investigation into the factors influencing surgical outcomes and complications in TR surgeries. This retrospective cohort study aims to investigate the occurrence of perioperative complications and their associated factors for TR of primary thoracic and lumbar spinal tumors over the past decade at our hospital, with a particular focus on the impact of tumor location on surgical success.

Materials and methods

Inclusion criteria

Patients diagnosed with primary thoracic and lumbar tumors who underwent to TR surgery at our hospital between 2010 and 2020 were included. All patients received a pathological diagnosis confirming primary spinal malignancy or a benign yet aggressive tumor. Inclusion criteria were: (1) pathologically diagnosed primary spinal tumors, (2) tumor located in the thoracic and lumbar spine, (3) TR surgery at our center, (4) regular follow-up, and (5) availability of all clinical data. Exclusion criteria were: (1) lost to follow-up, (2) lack of pathological diagnosis, and (3) coexistence with other malignancies.

Surgical techniques

Our cohort underwent three subgroups of TR procedures.

#1 TER, preserving a tumor-free margin, was performed following Boriani and colleagues' proposals [4, 5]. Boriani et al. recommended a detailed classification (Weinstein-Boriani-Biagini, WBB) system, and surgical approach for en-bloc resection of spinal tumors [4]. Firstly, we categorized the spine into various anatomical zones, allowing for precise planning based on tumor location and extent. Typically, TER involves a posterior or combined anterior–posterior approach, depending on the tumor's WBB subtypes and its relation to critical structures. This procedure ensures the complete removal of the tumor as a whole, even in cases involving the pedicle, eliminating the need for intralesional resection;

#2 some cases underwent TES using a T-saw [2, 3]. This method involves a meticulous dissection of the affected vertebra, followed by the controlled cutting of the bone using a T-saw, which allows for a clean separation of the tumor from surrounding tissues. The T-saw provides a controlled and minimally invasive means of cutting through the vertebral body, including the pedicles, thereby minimizing the risk of tumor spillage and enhancing the likelihood of achieving an R0 resection. TES is typically performed through a posterior approach;

and #3 others underwent gross total resection via a piecemeal approach. This approach is often used when en-bloc resection is not feasible due to the tumor’s size, location, or involvement of critical structures. The technique includes careful dissection around the tumor to minimize damage to surrounding tissues, followed by gradual removal of the tumor in a controlled manner.

After the removal of the affected vertebral segments, anterior column reconstruction was carried out using either titanium mesh or a 3D-printed artificial vertebral body. Surgical preparation included lesion and patient assessment, computed tomography (CT)-guided or open biopsy, blood preparation, and, when necessary, embolization of tumor-feeding vessels. The utilization of an ultrasonic bone scalpel commenced in late 2016.

Data collection

The collected dataset encompassed clinical features, surgical information, perioperative events, management and results. Neurological function was evaluated based on Frankel scales. Complications were evaluated using McDonnell’s methods [12] and categorized into major and minor subgroups based on whether they significantly impacted the patient’s recovery process, prolonged hospital stay, or increased medical costs. The primary outcome index was the complication rate.

Patient privacy was safeguarded through the anonymization of imaging and clinical information. Approval and supervision of this retrospective study were obtained from our institutional ethics committee board, and informed consent was acquired from all participants.

Statistical analysis

The Lilliefors test, an adaptation of the Kolmogorov–Smirnov test, determined the normal distribution of data. Data were presented as n (percentages), mean (standard deviation), or median (interquartile range). Statistical comparisons between groups were conducted using Pearson’s χ2 test (or Fisher’s exact test), unpaired student t test, or Mann–Whitney U test, as appropriate. Binary logistic regression analysis identified perioperative complication risk factors. SPSS 23 software (IBM, USA) facilitated data analysis, and a significance level of P < 0.05 was adopted.

Results

Demographics and clinical features

A total of 103 patients were included. The mean age of the cohort was 35.8 years (Table 1). 92 cases (89.3%) presented with local pain preoperatively, while 34 cases (33.0%) exhibited neurological deficit symptoms. Three cases (2.9%) were asymptomatic but detected through routine health examinations. Preoperative neurological function revealed 69 patients (67%) graded as Frankel E, 25 (24.3%) as D, 6 (5.8%) as C, and 3 (2.9%) as A or B. Preoperative imaging indicated thoracic vertebra lesions in 73 cases (70.9%), lumbar vertebra lesions in 27 cases (26.2%), and thoracolumbar lesions in three cases (2.9%). Cord compression was evident in 51 cases (49.5%). All lesions invaded soft tissues surrounding the vertebral body, namely WBB layer A and/or D.

Table 1.

Clinical features of the cohort (n = 103)

Items	Values
Male/female, n (%)	53 (51.5%)/50 (48.5%)
Age (years), mean (SD)	35.8 (14.2)
Hospitalization days, mean (SD)	21.7 (13.0)
Symptoms, n (%)
Local pain	92 (89.3%)
Neurological dysfunction	34 (33.0%)
Asymptomatic	3 (2.9%)
Frankel grades, n (%)
Frankel A	1 (1.0%)
Frankel B	2 (1.9%)
Frankel C	6 (5.8%)
Frankel D	25 (24.3%)
Frankel E	69 (67%)
Location of tumor, n (%)
Thoracic	73 (70.9%)
Lumbar	27 (26.2%)
Thoracolumbar	3 (2.9%)
Pathological subtypes, n (%)
Malignant	34(33.0%)
Benign aggressive	69 (67.0%)

SD Standard deviation, IQR Interquartile range

Surgical data

The median number of removed vertebral levels was 1 (2) (Table 2). Of the cases, 84 (81.6%) underwent total removal of the involved vertebrae (Fig. 1), while 19 (18.4%) underwent sagittal vertebral resection (Fig. 2). For the anterior column reconstruction, customized 3D-printed prosthesis was implanted in 49 cases (47.6%), titanium mesh in 51 cases (49.5%), and no implant was used between the vertebral bodies in the remaining 3 cases (2.9%).

Table 2.

Surgery-related data of the cohort (n = 103)

Items	Values
Tumor-feeding vessel embolization, n (%)	54 (52.4%)
Surgical approach, n (%)
Anterolateral alone	2 (1.9%)
Posterior alone	49 (47.6%)
Combined approach, one-staged	26 (25.2%)
Combined approach, two-staged	26 (25.2%)
Resected extent, n (%)
Total spondylectomy	84 (81.6%)
Sagittal vertebral resection	19 (18.4%)
Resected vertebral levels, median (IQR)	1 (2)
Surgical duration (min), mean (SD)	527 (206)
Estimated blood loss (ml), mean (SD)	2012 (1365)

SD Standard deviation, IQR Interquartile range

Fig. 1.

Illustrative case of total en-bloc resection. A 71-year-old male with L1 giant cell tumor underwent a single posterior surgical approach. a and b reveal the tumor invades vertebral body and right pedicle. c and d showcase the en-bloc resected tumor and the L1 vertebral body specimen. e is the postoperative X-ray

Fig. 2.

Illustrative case of sagittal vertebral resection. A 34-year-old male with L1-3 chondrosarcoma underwent combined retroperitoneal-posterior surgical approach. Preoperative images of the tumor are depicted in pictures a-c; pictures d-g illustrate the specimen and its corresponding X-rays; pictures h and i showcase postoperative X-rays

Single anterolateral or posterior approaches were taken in 49.5% (51/103) of cases, while combined anterior/anterolateral and posterior approaches were adopted in 50.5% (52/103). The average surgical time was 527 (206) minutes, with a mean estimated blood loss of 2012 (1365) ml. Embolization of tumor-supplying arteries was performed in 54 cases (52.4%). However, there was no significant difference in blood loss between patients who underwent preoperative arterial embolization and those who did not (P > 0.05).

Pathological distribution

Of the cases, 34 were malignant tumors, and 69 were benign aggressive tumors (Fig. 3). Benign tumors included 41 cases of giant cell tumor (39.8%) and 12 cases of osteoblastoma (11.7%), while chondrosarcoma (11.7%) was the most common malignancy.

Fig. 3.

Pathological subtypes of the whole cohort

Overall perioperative complications

A total of 79 patients (76.7%) experienced one or more perioperative complications, resulting in 161 complication events (Fig. 4), with an average rate of 1.56 per patient for the cohort. According to McDonell’s proposal, there were 71 (71/161, 44.1%) events graded as the major complications and 90 (90/161, 55.9%) as the minor. Major complications occurred in 46 patients (44.7%), and minor complications occurred in 33 patients (32.0%). No death was recorded during the indexed hospital stay. Mean hospital stay days differed significantly between patients with or without major complications (16.9 versus 27.6, P < 0.001).

Fig. 4.

Presentation of 161 complication events, with chest complications, neurological deterioration, and pleural rupture being the most common

Intraoperative complications

Seven patients (6.8%, 7/103) experienced significant vascular injuries. The distribution included: one involving the inferior vena cava, four affecting the iliac vein, one involving the azygos vein, and the last one affecting the segmental artery and aorta. Notably, three injuries occurred during the posterior approach, three during the anterolateral approach, and one during the anterior approach.

Five cases underwent immediate vascular suture, one achieved hemostasis through compression, and one required a transfer to a transabdominal approach for direct vessel suturing. All repair procedures resulted in satisfactory outcomes.

Dural tears were observed in 17.5% (18/103) of cases, and pleural rupture occurred in 20.4% (21/103) of cases. The majority of patients with these complications underwent suture repair.

Early postoperative complications

Immediate neurological deterioration occurred in 22.3% (23/103) of cases (Fig. 4). Among these cases, 22 patients (21.4%) manifested decreased slight-to-moderate muscle strength or hypoesthesia, while one case developed paraplegia. In response, all patients, with the exception of one, underwent conservative treatment involving intravenous methylprednisolone and other medications. Encouragingly, 91.3% (21/23) of these cases achieved neurological recovery before discharge.

Chest complications were noted in 29.2% (30/103) of cases, encompassing pleural effusion and pneumothorax. Pleural injury or planned transthoracic approaches resulted in a significantly higher rate of chest complications compared to patients with intact pleura (51.5% versus 18.6%, P < 0.001). Respiratory complications occurred in 13.6% (14/103) of cases, all of whom recovered after anti-infection treatment and oxygen inhalation.

Wound healing problems were observed in 4.9% (5/103) of cases, and four of them underwent reoperation. Other complications, such as cardiovascular, urinary, and stroke, were also noted (Fig. 4). Deep venous thrombosis (DVT) in the lower extremity was identified in three patients (2.9%), all of whom achieved stabilization through immobilization and anti-coagulation treatments, with no instances of pulmonary embolism. Three patients (2.9%) exhibited internal fixation malposition identified through imaging examination, necessitating reoperation for hardware adjustment (Fig. 5). Two patients (1.9%) developed postoperative chylous leakage, which improved with long-term drainage, nutritional support, diet control, and drug treatment.

Fig. 5.

Illustrative case of hardware malposition. 55-year-old male with giant cell tumor at the thoracic spine. a and (b), preoperative CT scans showing osteolytic lesions at T6; c and (d), preoperative MR images demonstrating the tumor primarily as vertebral destruction and large paravertebral mass, with no spinal cord compression; e, CT image 3 days postoperatively revealing posterior displacement of the titanium mesh; f, CT image showing satisfactory repositioning of the titanium mesh via re-operation

Complications-associated factors

A total of 54 patients (52.4%) underwent single-level vertebral resection, while 49 patients (47.6%) underwent multilevel resection (Table 3). The incidence of complication events per patient significantly differed between the two groups (P = 0.031). Specifically, there were variations in the incidence of respiratory, chest, and hematological complications (anemia and DVT) (P < 0.05).

Table 3.

Comparison of complications between single-level and multi-level vertebral resection

	Single-level group (n = 54)	Multi-level group (n = 49)	P values
Age (years), mean (SD)	35.9 (14.1)	35.8 (14.1)	0.984
Events per patient, mean (SD)	1.26 (1.07)	1.92 (1.62)	0.031*
Surgical duration (min), mean (SD)	453 (187)	616 (190)	< 0.001*
Estimated blood loss (ml), mean (SD)	1771 (1051)	2284 (1594)	0.189
Hospital stay (days), mean (SD)	17.0 (6.7)	26.9 (15.9)	< 0.001*
Large vascular injury, n (%)	6 (11.1%)	1 (2.0%)	0.151
Dural tear, n (%)	8 (14.8)	10 (20.4)	0.455
Pleural injury, n (%)	14 (25.9)	7 (14.3)	0.143
Neurological deterioration, n (%)	12 (22.2%)	11 (22.4%)	0.978
Cerebrospinal fluid leakage, n (%)	11 (20.4%)	7 (14.2%)	0.417
Respiratory complications, n (%)	2 (3.7%)	12 (24.5%)	0.002*
Chest complications, n (%)	4 (7.4%)	26 (53.1%)	< 0.001*
Intestine infection, n (%)	1 (1.6%)	0	1.000
Wound healing problems, n (%)	1 (1.6%)	4 (8.2%)	0.303
Hematological complications, n (%)	3 (5.6%)	9 (18.4%)	0.043*
Hardware malposition, n (%)	1 (1.6%)	2 (4.1%)	0.932
Cardiovascular complications, n (%)	0	1(2.0)	0.476
Intracranial hemorrhage, n (%)	1 (1.6%)	1 (2.0%)	1.000
Urinary complications, n (%)	3 (5.6%)	0	0.244
Chylous leakage, n (%)	0	2 (4.1%)	0.224

Comparison was conducted utilizing unpaired student t test, Mann–Whitney U test, Chi-squared test or Fisher’s exact test accordingly

^*Significantly different at P < 0.05

In the lumbar tumor group, the overall complication events per patient were comparable to those in the thoracic group (P = 0.404, Table 4). However, the lumbar group exhibited significantly higher estimated blood loss (P = 0.035), longer duration of surgery (P = 0.019), and increased rates of vascular injury (P = 0.001), postoperative neurological deterioration (P = 0.003), and CSF leakage (P = 0.001) compared to the thoracic vertebrae.

Table 4.

Comparison of complications between thoracic tumor versus lumbar tumor groups

	Thoracic group (n = 73)	Lumbar group (n = 27)	P value
Events per patient, mean (SD)	1.41 (1.28)	1.67 (1.32)	0.404
Surgical duration (min), mean (SD)	485 (174)	600 (235)	0.019*
Estimated blood loss (ml), mean (SD)	1859 (1335)	2409 (1375)	0.035*
Large vascular injury, n (%)	1 (1.4)	6 (22.2)	0.001*
Neurological deterioration, n (%)	10(13.7)	11(40.7)	0.003*
Chest complications, n(%)	27 (37.0)	2 (7.4)	0.004*
Respiratory complications, n(%)	12 (16.4)	2 (7.4)	0.406
Cerebrospinal fluid leakage, n(%)	6 (8.2)	10 (37.0)	0.001*
Internal fixation failure, n (%)	2 (2.7)	1 (3.7)	> 0.999

Comparison was conducted utilizing unpaired student t test, Mann–Whitney U test, Chi-squared test or Fisher’s exact test accordingly

^*Significantly different at P < 0.05

Compared with patients undergoing inaugural treatment, recurrent cases required a higher number of resected vertebral segments (P = 0.011, Table 5). Patients with recurrent tumors had a higher incidence of overall perioperative complications per patient (P = 0.011). The rate of dural tear significantly increased in recurrent cases (P = 0.002). Compared to benign tumors, primary malignant tumors were associated with a longer surgical duration (P = 0.005), extended hospital stays (P = 0.009), and a higher incidence of wound healing problems (P = 0.04, Table 6).

Table 5.

Comparison of complications between primary and recurrent cases

	Primary (n = 86)	Recurrent (n = 17)	P-value
Age (years), mean (SD)	35.3 (13.6)	38.6 (18.6)	0.382
Events per patient, mean (SD)	1.37 (1.29)	2.53 (1.66)	0.011*
Resected vertebral segments, median (IQR)	1 (2)	2 (1)	0.011*
Surgical duration (min), mean (SD)	504 (197)	649 (219)	0.015*
Estimated blood loss (ml), mean (SD)	1975 (1363)	2216 (1408)	0.518
Hospital stay (days), mean (SD)	20.9 (13.3)	25.5 (11.5)	0.191
Large vascular injury, n (%)	6 (7.0%)	1 (5.9%)	> 0.999
Dural tear, n (%)	10 (11.6%)	8 (47.1%)	0.002*
Neurological deterioration, n (%)	17 (19.8%)	6 (35.3%)	0.277
Cerebrospinal fluid leakage, n (%)	13 (15.1%)	5 (29.4%)	0.285
Respiratory complications, n (%)	10 (11.6%)	4 (23.5%)	0.357
Chest complications, n (%)	22 (25.6%)	8 (47.1%)	0.137
Wound healing problems, n (%)	4 (4.7%)	1 (5.9%)	> 0.999

SD stands for standard deviation, IQR interquartile range

Comparison was conducted utilizing unpaired student t test, Mann–Whitney U test, Chi-squared test or Fisher’s exact test accordingly

^*Significantly different at P < 0.05

Table 6.

Comparison of complications between malignant and benign cases

	Malignant (n = 34)	Benign (n = 69)	P
Resected vertebral segments, median (IQR)	2 (2)	1 (2)	0.133
Complication rate, n (%)	26 (76.5)	53 (76.8)	0.969
Surgical duration (min), mean (SD)	610 (236)	482 (174)	0.005*
Estimated blood loss (ml), mean (SD)	2023 (1421)	2008 (1348)	0.758
Hospital stay (days), mean (SD)	26.4 (17.8)	19.4 (9.2)	0.009*
Wound healing problems, n (%)	4 (11.8)	1 (1.4)	0.040*

SD Stands for standard deviation, IQR Interquartile range

Comparison was conducted utilizing unpaired student t test, Mann–Whitney U test, Chi-squared test or Fisher’s exact test accordingly

^*Significantly different at P < 0.05

Univariate analysis identified factors associated with major perioperative complications (Table 7), including age older than 35 years (P = 0.048), embolization of tumor-feeding vessels (P = 0.036), multilevel resection (P < 0.001), and combined approaches (P = 0.005). Subsequent binary logistic stepwise regression analysis indicated that multilevel resection (P < 0.001) was independent risk factor for major perioperative complications.

Table 7.

Binary logistic regression analysis of perioperative major complication risk factors

		Univariate regression analysis		Multivariate regression analysis
		OR (95%CI)	P values	OR (95%CI)	P values
Age	> 35y	2.229(1.008–4.927)	0.048*	2.530(0.990–6.470)	0.053
	≤ 35y (ref.)
Gender	Male	0.811(0.372–1.767)	0.598
	Female (ref.)
Preoperative spinal cord compression	Yes	0.885(0.407–1.926)	0.758
	No (ref.)
Preoperative embolization	Yes	2.346(1.057–5.204)	0.036*	2.088(0.834–5.229)	0.116
	No (ref.)
Number of resected levels	Multi-level	6.505(2.743–15.426)	< 0.001*	5.938(2.281–155.459)	< 0.001*
	Single level (ref.)
Recurrent tumor	Yes	2.671(0.904–7.896)	0.076
	No (ref.)
Approach	Combined	3.214(1.428–7.233)	0.005*	2.058(0.809–5.235)	0.13
	Single (ref.)

OR Odds ratio, CI Confidential interval

^*Significant at P < 0.05

Discussion

This study adds valuable insights into the perioperative complications associated with total spinal tumor resection surgery, particularly in the thoracic and lumbar regions. The large sample size contributes to the robustness of the findings, setting it apart from other studies in the field. We conducted an in-depth exploration on this incidence, their subtypes, and associated factors. To our knowledge, this article is unique in addressing this specific aspect of spinal surgery.

In our research, we observed that 76.7% of patients experienced at least one perioperative complication event. This rate is consistent with findings from previous studies (Table 8). Demura et al. [9] and Bandiera et al. [11] and reported complication rates of 50.0% and 67%, respectively, following en-bloc resection of spine tumors. Smaller-scale studies reported complication rates ranging from 52% to 86.7% [8, 13–16]. The higher complication rate observed in this cohort underscores the complexity and challenges associated with total spinal tumor resection.

Table 8.

Summary of previous articles on complications following total resection of spinal tumors

Authors	Year of publication	Sample size	Pathological subgroups	Surgical procedures	Complication rates (%)
Matsumoto et al. [8]	2015	33	Primary and metastatic	TES	72.7
Luzzati et al. [15]	2015	38	Primary and metastatic	TES	65.8
Boriani et al. [6]	2016	220	Primary and metastatic	TER	46.2
Sciubba et al. [16]	2016	23	Primary	TES	65.2
Shah et al. [7]	2017	33	Primary and metastatic	TES	51.5
Shimizu et al. [9]	2018	30	Primary	TES	86.7
Araujo et al. [14]	2018	17	Primary and metastatic	TES	76.5
Demura et al. [10]	2021	307	Primary and metastatic	TES	67.1
Tang et al. [13]	2021	27	Primary and metastatic	TES	55.6
Bandiera et al. [11]	2022	298	Primary and metastatic	TER	50.0

TES Stands for total en-bloc spondylectomy, TER Total en-bloc resection

The most severe intraoperative complication identified in this study was large vascular injury, with a significantly higher risk observed in lumbar tumor surgery compared to thoracic surgery (Table 4). Notably, the iliac vein was found to be the most susceptible site, attributed to local anatomical features. Caution is recommended, particularly during posterior surgery, and an auxiliary retroperitoneal or transabdominal approach may mitigate the risk of large vessel injury. Our findings align with the recommendation by Kawahara et al.[17], advocating for a combined retroperitoneal/transabdominal-posterior approach for tumors in L4 or L5 vertebrae. Therefore, we recommend combined retroperitoneal/transabdominal-posterior approaches for tumors residing in layer A, sectors 6–8 by WBB system, especially those with huge mass anterior to the lower lumbar. Another advantage is to potentially reduce neurological injury risk in the combined approach (Table 7).

In our cohort, 52.4% of patients underwent preoperative embolization of tumor-feeding vessels within 48 h before surgery. All these patients underwent this procedure without complications, highlighting this is a safe procedure. Embolization proved beneficial in reducing intraoperative blood loss, shortening operation time, providing a clear surgical field, and improving tumor resectability [18, 19]. Usually, we examined the vascularity status of the target lesions using digital substraction angiography, and only performed the tumor-feeding vessel embolization for the hypervascular ones. This finding emphasizes the importance of tailored and vigilant postoperative strategies to mitigate and treat specific complications, contributing to improved patient outcomes in the complex field of spinal tumor surgeries.

Postoperative neurological deterioration affected 22.3% of patients in our study, while Shimizu et al. [9] reported lower limb muscle strength decline in up to 80% of patients after en-bloc resection of lumbar tumors. The high risk of postoperative neurological deterioration is assumed to be associated with inappropriate traction of nerve roots during the operation. The subsequent amelioration of this complication was accomplished through intravenous methylprednisolone and other drug therapies for most of our patients.

CSF leakage occurred in 17.5% of patients, predominantly with a favorable prognosis after conservative treatment. Recurrent cases were more prone to CSF leakage, emphasizing the need for special attention and potential debridement surgery in cases with poor wound healing or prolonged exudation. In addition, we noticed that the use of the ultrasonic bone scalpel offered significant advantages in safety for cutting the lamina and facet structures, as it could reduce operative time and minimize blood loss. However, the advantages of using the ultrasonic bone scalpel for vertebral body and pedicle resection compared to the T-saw require further studies to establish a clearer comparison of safety and efficiency.

The frequent occurrence of chest cavity and/or respiratory complications poses significant challenges in the postoperative care of patients undergoing spinal tumor resection, particularly those with thoracic tumors (Table 4). These complications have been observed to substantially prolong hospital stays, including potential extended periods in the intensive care unit (ICU). The inherent complexity of thoracic tumor cases often necessitates surgical approaches involving transthoracic procedures or even partial pleural and rib removal to achieve en-bloc resection. The additional complications of CSF and chylous leakage further exacerbates the situation. In light of these complexities, proactive measures such as prophylactic chest drainage become imperative for patients undergoing spinal tumor resections at the thoracic region.

While this retrospective study contributes valuable insights into the safety of total spinal tumor resection during the perioperative and short-term postoperative stages, it acknowledges the following limitations. The focus on short-term outcomes and lack of long-term follow-up data represent potential limitations. Additionally, the study's exclusive consideration of TR in the thoracic and lumbar spine, excluding cervical tumors or solitary metastatic lesions, is acknowledged due to the unique challenges and complexities associated with these cases [20–23].

Conclusion

This study provides a comprehensive analysis of perioperative complications in patients undergoing total resection surgery for thoracic and lumbar tumors. The overall incidence came up to 76.7%. Neurological deterioration, chest/respiratory events, and CSF leakage were the most common moderate-to-severe complications. Other severe or even life-threatening events, including DVT, large vessels injury, and cardiocerebral vascular events sometimes occurred. Factors, such as old age, surgical segments, extent of resected vertebral levels, and surgical approaches were identified as related to the occurrence of complications. The study underscores the importance of a high patient-volume center, a multidisciplinary team, and meticulous surgical preparation to enhance the safety of total resection surgery.

Authors’ contributions

JL and PH designed the study, reviewed the patients, collected and processed the clinical data, drafted the manuscript, and presented the attached figures. FW designed the study, selected the patients, acquired funding support, reviewed & edited the manuscript, and supervised the study. HZ reviewed the patients, collected the data and presented the attached figures. XL and ZL provided methodologic support, reviewed & edited the manuscript, and supervised the study. All authors reviewed and approved this manuscript.

Funding

This study was supported by Peking University Third Hospital (grant number: BYSYZD2022023, P. Hu; Y73504-10, F. Wei). The fund provider has no influence on the design of the study and data collection, analysis, and interpretation of data and in the preparation of the manuscript.

Data availability

Data is not publicly available due to containing patient privacy. Data may be obtained from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the institutional ethics committee of Peking University Third Hospital (approval number: M2023797), and performed in accordance with the ethical standards of the Declaration of Helsinki (1964) and its subsequent amendments.

Competing interests

The authors declare no competing interests.