Cancer is one of the leading causes of death worldwide, with most patients dying from metastatic disease.1 It is currently estimated that more than 60% of patients with cancer will require surgery for the removal of solid tumours.2 The processes of tissue trauma, surgical manipulation of the tumour, and exposure to the physiologic stresses of the perioperative period can result in impaired local and cellular immunity, with consequent loco-regional recurrence and metastasis.3,4,5 The degree to which general anesthetic technique (inhalational volatile agents, or total intravenous anesthesia [TIVA] with propofol) contributes to this patient vulnerability in the perioperative period is an area of particular interest.5

Preclinical studies have found that both intravenous and volatile anesthetic agents alter the biology of cancer and immune cell lines by directly activating cellular receptors and cell signaling pathways, as well as by altering cellular kinetics and gene transcription.3,6,7 Anesthetic technique may also affect cell-mediated immunity and promote spread in different cancer types.8

A recent systematic review published by Soltanizadeh et al. examined outcomes of cancer surgery after inhalational vs intravenous anesthesia, but was constrained by the inclusion of studies that focused on postoperative complications or studies that did not have cancer recurrence or survival as the primary endpoint.9 We conducted this meta-analysis to provide an up-to-date assessment of the current evidence for the impact of type of anesthesia for cancer surgery on long-term clinical outcomes (cancer recurrence and overall survival).

Methods

Our results are reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA guidelines) and this review was registered on 8 October, 2018 with PROSPERO (registration number: CRD42018081478).

Study eligibility criteria

All randomized-controlled trials (RCTs) and observational longitudinal studies (prospective or retrospective) evaluating the effects of TIVA and inhalational anesthetic agents on cancer outcomes in patients undergoing cancer surgery were included. We excluded animal studies, studies not published in English, studies with insufficient information to perform the meta-analysis (such as no cancer-related endpoint), studies that did not report separate data for each intervention group with a measure of effect (such as hazard ratio [HR] or Peto odds ratio), or studies that focused on other interventions (e.g., perioperative chemotherapy and/or radiation therapy).

Information sources

The databases initially searched were Medline (through Ovid), EMBASE (through Ovid), The Cochrane Library, Web of Science, and PubMed from the start of inception until 17 March 2017, with a search update performed in PubMed and in sources of grey literature until 14 November 2018. Sources of grey literature included Open Grey and Google Scholar®. Conference proceedings and abstracts were searched in Web of Science. EndNote software (Clarivate Analytics, Philadelphia, PA, USA) was used to store all citations for duplicate checking.

Search

An experienced librarian (G.P.) developed a comprehensive search strategy that included broad terms such as “cancer”, “tumor”, “neoplasms”, “perioperative”, “anesthesia”, and narrow terms such as “TIVA”, “propofol”, “volatile”, and “sevoflurane” among other anesthetic intervention terms. The complete Medline search strategy and the terms for the search update are reported in Appendix 1.

Study selection

Eligibility assessments were performed independently by three teams of reviewers on behalf of the Global Onco-Anesthesia Research Collaboration Group (Acknowledgment). Cohen’s kappa coefficient, which describes the level of inter-rater agreement, was calculated. Disagreements at all stages were resolved through discussion. If agreement could not be reached, a third reviewer (B.R.) made a final decision.

Data collection process and data items

One reviewer (A.Y.) extracted data from individual studies and another reviewer (M.L.O.) cross-checked the information. The following information was extracted from each study: i) general information such as title, authors, publication year, and country; ii) study characteristics such as study design, setting, sample size, and outcome assessed; iii) participant characteristics such as number of patients within each group, age (mean and range) and type of cancer; and iv) intervention characteristics such as details of anesthetic techniques (volatile agents used, TIVA agent used), concomitant drug (opiates, anti-inflammatories and blood transfusions), or regional anesthesia use.

Risk of bias in individual studies

One reviewer (A.Y.) assessed the risk of bias of the included studies and another reviewer (B.R.) cross-checked the information. For RCTs, we used the Cochrane risk of bias tool. The potential of bias was appraised in five domains: selection, performance, detection, attrition, and reporting. These domains specifically evaluate how the random sequence was generated, methods of allocation concealment, blinding of participants and personnel, blinding of the outcome assessment, how incomplete outcome data were handled, and if there was evidence of selective outcome reporting. Each potential source of bias was graded as low, unclear, or high, and a justification for each judgment was provided. Observational studies were evaluated using the Newcastle Ottawa Scale (NOS),10 which assesses the potential for bias by scoring the selection process of the study groups, comparability of the groups, and ascertainment of exposure and outcome in the studies. Studies can be awarded a maximum of one point for each domain with an additional point being awarded for studies controlling for additional confounders. The maximum score allocated in the selection domain is 4 points, in the comparability domain is 2 points and outcome domain is 3 points. A maximum score of 9 points can be achieved and a higher score (≥ 7 points) indicates a lower risk of bias.11

Summary measures

The primary outcome measures for this meta-analysis were recurrence-free survival and overall survival. Adjusted HR from Cox proportional hazard models, their respective 95% confidence intervals (CI), and the P values were extracted from each of the studies. Where more than one data set was given, multivariate analysis data were used. If this was not provided, propensity matched data, if available, were used instead.

Synthesis of results

Analyses were conducted using Review Manager (RevMan version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). We converted the reported HRs into log HRs and used the generic inverse-variance method with random-effects model to pool the data. To maintain TIVA as a reference group across studies, we inverted the HR (1/HR) for studies reporting volatiles as a reference group. When data were provided in Kaplan Meier plots, an attempt was made to contact the study authors for further data. When the estimates differed substantially among the pooled studies (Chi squared test, P = 0.10), we conducted a sensitivity analysis by eliminating the outliers.

Additional analyses and risk of bias across studies

We a priori planned to explore sources of heterogeneity with subgroup analyses or meta-regression as well as using a funnel plot and a regression asymmetry test to assess small-study bias. Because of the small number of studies included, this was not done.

Results

Study selection

Figure 1 shows the study selection flowchart. We retrieved 12,508 citations and after removal of duplicates, there were 9,536 unique citations. After review, one prospective RCT and nine retrospective studies that pertained to volatile anesthesia and propofol-TIVA were included in the final analysis.12,13,14,15,16,17,18,19,20,21

Fig. 1
figure 1

PRISMA flow diagram of study selection

Full size image

Study characteristics and risk of bias within studies

Table 1 shows the study characteristics and Table 2 the recorded event rates (cancer recurrence or death). All the nine retrospective studies had a NOS ≥ 7 demonstrating good methodologic quality and a low risk of bias.12,13,14,15,16,17,18,20,21 The single prospective RCT included in this meta-analysis had a low risk of bias for each domain.19

Table 1 Characteristics of each study
Full size table
Table 2 Number of patients with events (cancer recurrences or death) reported
Full size table

Results of individual studies

Patients receiving a propofol infusion only, or a propofol and remifentanil infusion during their surgery were categorized into the TIVA group; patients receiving sevoflurane, isoflurane, desflurane or enflurane were categorized into the volatile group. All studies except Yan et al.19 adjusted for at least one of the following variables in their multivariate analyses: age, body mass index, comorbidities, preoperative therapy, pathologic stage or grade of cancer, and intraoperative anesthetic interventions such as epidural or blood transfusion.12,13,14,15,16,17,18,20,21

Synthesis of results

Recurrence-free survival

Six studies (five retrospective13,14,15,16,20 and one RCT)19 examined the effects of TIVA and volatile agents on recurrence-free survival in breast, esophageal, and non-small cell lung cancer (Fig. 2). The total sample size was 7,866 patients. When compared with volatile anesthesia, the use of TIVA was associated with improved recurrence-free survival in these cancer types (pooled HR, 0.78; 95% CI, 0.65 to 0.94; P < 0.01).

Fig. 2
figure 2

Recurrence-free survival

Full size image

Overall survival

Eight studies (seven retrospective12,13,16,17,18,20,21 and one RCT),19 that included a total of 18,778 patients, provided ten HRs for this analysis (Fig. 3) in breast, colorectal, gastric, esophageal, and non-small cell lung cancer, and mixed cancer types. There was an associated improvement in overall survival with TIVA use when compared with volatile anesthesia (pooled HR, 0.76; 95% CI, 0.63 to 0.92; P < 0.01).

Fig. 3
figure 3

Overall survival

Full size image

There was substantial heterogeneity among the studies and to explore it, outliers were removed (four estimates: Jun et al.,13 Wigmore et al.,17 Wu et al.18 and Zheng et al.).21 This gave an inconsistency score of 0% and a resulting pooled HR of 0.97 (95% CI, 0.85 to 1.11; P = 0.66). Unsuccessful strategies to further decrease the inconsistency score included removing the only RCT, removing the one study with multiple estimates (Enlund et al.),12 and leaving studies with positive estimates only.

Discussion

Despite advances in modern medicine, cancer is still a leading cause of death worldwide.22 It is therefore vital that clinicians consider all aspects of cancer care, including the delivery of anesthesia during cancer resection surgery, to optimize patients’ cancer outcomes. The pooled results from this meta-analysis suggest that TIVA use (compared with volatile anesthesia) during cancer surgery is associated with improved recurrence-free survival and overall survival across numerous cancer types. Breast cancer was the most often examined tumour type with five studies reporting on outcomes after breast cancer surgery12,14,15,19,20; in this population, TIVA use was associated with an improvement in recurrence-free survival but not overall survival.

The inconsistency in results between the individual studies included within the meta-analysis for breast cancer outcomes may be explained by confounding factors such as the degree of surgical trauma.23 In the study by Kim et al., where no benefit was reported with TIVA use across all types of breast cancer surgery, it is important to note that of those patients who had suffered recurrence, 73% of patients had undergone mastectomy.14 Similarly, in the study by Yoo et al., patients undergoing total mastectomy were associated with higher risks of cancer recurrence and all-cause mortality when compared with breast-conserving surgery.20 In the study by Lee et al., all of the patients underwent modified radical mastectomies and had a lower rate of cancer recurrence with TIVA use when compared with the sevoflurane group.15 Enlund et al. examined overall survival in patients with breast cancer but did not specify the type of surgical procedure.12 Importantly, the five-year survival rate for breast cancer is 88.9%, and thus the 50-70 month follow-up time in this study by Enlund et al.12 may not be sufficient to detect a meaningful difference of the effect of anesthesia type on survival.24

Preclinical studies suggest that drugs used for general anesthesia affect cellular immunity and potentiate cancer spread.3,25 Mechanistic studies have examined the differential effects of anesthetic agents on tumour cell biology, with in vitro data strongly supporting a pro-metastatic effect of volatile anesthesia and an anti-metastatic effect of propofol.3In vitro studies investigating the effect of different volatile agents have found an increased expression of cellular mediators that promote cancer cell proliferation, resistance of apoptosis by tumour cells, a propensity to invasion and migration of cells, endothelial-mesenchymal transition, basement membrane degradation, and angiogenesis.26,27,28,29,30,31 In contrast, when tumour cells are exposed to propofol, apoptosis is preserved and cell proliferation is reduced.32,33,34,35

Volatile anesthesia’s alteration of immune function has also been implicated in its hypothesized pro-metastatic potential through manipulation of the perioperative immune response.3 Preclinical data have reported impaired immune cell number and function after exposure to volatile anesthesia in animal models of cancer.3,25,36,37 Volatile agents reduce natural killer cell activity, a cytotoxic lymphocyte in the innate immune system and critical in the anti-tumour immune response.36,38,39 Reduced natural killer cell activity has been linked to tumour cell dissemination in patients with cancer.40,41,42 In contrast, in vitro studies report that propofol does not affect natural killer cell activity.39 Propofol may also reduce hypoxia-inducible factor 1α (HIF-1α) levels, a key regulator in the response to tumour growth.30 Activation of HIF-1α occurs during low oxygen states and promotes cell proliferation, angiogenesis, and metastasis43; this has been reported to be activated by volatile agents.26,30,44

It is therefore plausible that anesthesia technique is a critical component in cancer progression. Volatile agents may potentially “fuel the fire” and contribute to inherent cancer and surgical wounding processes characterized by pro-adrenergic, pro-inflammatory, immunomodulatory, and pro-angiogenic signalling.5 No single pathway, however, has been implicated, suggesting heterogeneity of the underlying drivers of cancer recurrence. Other clinically relevant interventions in the perioperative period, including surgical extent, blood transfusion, hypothermia, and administration of other medications (e.g., opioids, beta-blockers, anti-inflammatories, steroids), may themselves impact cancer cell biology.5

Surgical trauma activates neuroendocrine, inflammatory, immunologic, and metabolic pathways.45 These changes reduce innate and cellular immunity and may promote cancer spread postoperatively.46 Postoperative complications, including wound complications, pulmonary infections, and anastomotic leaks, have been reported to increase cancer recurrence and reduce overall survival.47,48 Such complications are characterized by exaggerated inflammatory processes. It is important to note that postoperative complications after cancer surgery have also been reported to be associated with anesthetic technique. De la Gala et al. noted a reduction in postoperative pulmonary complications and one year mortality with sevoflurane use (compared with, TIVA) in patients undergoing lung resection surgery.49 Conversely, Chang et al. noted fewer pulmonary complications and reduced mortality with TIVA use in patients with head and neck cancer undergoing free flap surgery when compared with volatile anesthesia.50 Postoperative complications may adversely affect postoperative recovery and reduce the ability to “Return to Intended Oncologic (adjuvant) Therapy” (RIOT) in the immediate postoperative period.51

Surveys of current clinical practice report that anesthesiologists generally have a preference for volatile anesthesia.52,53 In a survey of Australasian anesthesiologists, Lim et al. reported that > 80% of anesthesiologists prefer volatile-based anesthesia within their daily routine. Despite 43% of respondents reporting that they felt that TIVA may reduce cancer recurrence (compared with, inhalational anesthesia), only 29% reported regular use of TIVA for cancer surgery.52 This propensity toward volatile-based anesthesia necessitates large prospective RCTs of TIVA vs volatile anesthesia to inform international clinical guidelines.

Limitations of this study include the retrospective nature of the majority of the studies. The studies also had different follow-up intervals and significant variability of baseline patient demographics. There were also differences in study characteristics, including variable sample sizes in the treatment arms,13,14 unbalanced study populations (e.g., patients in one treatment group being older, having significant comorbidities),18 different stages/grades of cancer, differences in anesthetic technique (e.g., remifentanil, different volatiles used, and difference in use of regional anesthesia), and differences in surgical technique (e.g., differences in surgical magnitude). This study was also limited by the availability of data within the published manuscripts for analysis. The possibility of publication bias (which could not be assessed because the overall number of included studies was too low) should also be considered as this has the potential to greatly affect the results of this meta-analysis. Given these limitations, while the results favour a positive impact of propofol-based TIVA on cancer outcomes, the data should be interpreted with caution.

Collectively, this meta-analysis examined over 21,000 cancer patients with multiple cancer types. Despite the heterogeneity of the study designs and data, including different cancer types, there is an association between improved cancer outcomes with propofol-based TIVA when compared with inhalational volatile-based anesthesia. The results of this meta-analysis, together with the growing body of preclinical literature in the field, support the hypothesis that choice of anesthetic drug may influence patient outcome after cancer surgery. To test the hypothesis, a number of prospective RCTs in specific cancer types are currently underway (Randomized, Open-label Study to Compare Propofol Anesthesia With Sevoflurane Anesthesia in Terms of Overall Survival in Patients With Surgical Intervention for either Breast-, Colon-, or Rectal Cancer [NCT01975064]; General Anesthetics in CAncer REsection Surgery [GA-CARES] Trial: Pragmatic Randomized Trial of Propofol vs Volatile Inhalational Anesthesia [NCT03034096]; Impact of Inhalational Versus Intravenous Anesthesia Maintenance Methods on Long-term Survival Rate in Elderly Patients After Cancer Surgery: an Open-label, Randomized-Controlled Trial [NCT02660411]; and Volatile Anaesthesia and Perioperative Outcomes Related to Cancer [VAPOR-C]: A Feasibility Study [ACTRN1261700106538]) and will help guide the optimal anesthesia choice for perioperative cancer care.