Effects of anesthesia on long-term survival in cancer surgery: A systematic review and meta-analysis

Backgrounds The association between anesthesia and long-term oncological outcome after cancer surgery remains controversial. This study aimed to investigate the effect of propofol-based anesthesia and inhalation anesthesia on long-term survival in cancer surgery. Methods A comprehensive literature search was performed in PubMed, Medline, Embase, and the Cochrane Library until November 15, 2023. The outcomes included overall survival (OS) and recurrence-free survival (RFS). The hazard ratio (HR) and 95 % confidence interval (CI) were calculated with a random-effects model. Results We included forty-two retrospective cohort studies and two randomized controlled trials (RCTs) with 686,923 patients. Propofol-based anesthesia was associated with improved OS (HR = 0.82, 95 % CI:0.76–0.88, P < 0.00001) and RFS (HR = 0.80, 95 % CI:0.73–0.88, P < 0.00001) than inhalation anesthesia after cancer surgery. However, these positive results were only observed in single-center studies (OS: HR = 0.76, 95 % CI:0.68–0.84, P < 0.00001; RFS: HR = 0.76, 95 % CI:0.66–0.87, P < 0.0001), but not in multicenter studies (OS: HR = 0.98, 95 % CI:0.94–1.03, P = 0.51; RFS: HR = 0.95, 95 % CI:0.87–1.04, P = 0.26). The subgroup analysis revealed that propofol-based anesthesia provided OS and RFS advantages in hepatobiliary cancer (OS: HR = 0.58, 95 % CI:0.40–0.86, P = 0.005; RFS: HR = 0.62, 95 % CI:0.44–0.86, P = 0.005), gynecological cancer (OS: HR = 0.52, 95 % CI:0.33–0.81, P = 0.004; RFS: HR = 0.51, 95 % CI:0.36–0.72, P = 0.0001), and osteosarcoma (OS: HR = 0.30, 95 % CI:0.11–0.81, P = 0.02; RFS: HR = 0.32, 95 % CI:0.14–0.75, P = 0.008) surgeries. Conclusion Propofol-based anesthesia may be associated with improved OS and RFS than inhalation anesthesia in some cancer surgeries. Considering the inherent weaknesses of retrospective designs and the strong publication bias, our findings should be interpreted with caution. Well-designed multicenter RCTs are still urgent to further confirm these findings.


Introduction
Cancer constitutes an enormous burden worldwide.It was estimated that there were 19.3 million new cancer cases globally and nearly 10.0 million cancer deaths in 2020.The worldwide cancer cases are predicted to increase to 28.4 million in 2040 [1].Surgical resection is considered the primary treatment for solid malignant tumors.However, cancer recurrence and metastasis after surgery still are the main reasons for cancer-related mortality [2].Accumulating evidence has shown that anesthetic drugs during surgery may disrupt the delicate balance between tumor invasiveness and host immune surveillance, then contribute to cancer recurrence and metastasis [3][4][5].Therefore, it is vital to explore the optimal anesthetics for patients undergoing cancer surgery.
Intravenous and inhalation anesthetics are two commonly used agents for general anesthesia.Studies have reported that inhalation anesthetics might exert suppressive effects on innate and adaptive immunity and enhance the metastatic abilities of cancer cells [6][7][8].On the contrary, propofol has been shown to preserve anti-tumor immunity and suppress the growth and survival of cancer cells [9][10][11][12].In addition, laboratory studies have demonstrated that exposure to inhalation anesthetics rather than propofol could induce the upregulation of hypoxia-inducible factor-1α and vascular endothelial growth factor, which promotes cancer cell progression [7,13,14].These data lead to the hypothesis that propofol might be superior to inhalation anesthetics in cancer surgery.Previous systematic reviews and meta-analyses investigated the association between intravenous and inhalation anesthetics and long-term survival of cancer patients [15][16][17][18], however, the conclusions remained inconsistent with insufficient sizes.Subsequently, more studies have been published to address this controversy [19][20][21][22][23][24].Therefore, we performed an updated systematic review and meta-analysis to investigate the effect of propofol-based anesthesia and inhalation anesthesia on long-term survival in patients undergoing cancer surgery.

Methods
This systematic review was conducted according to the methodology described in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Supplementary Table 1) [25].The protocol was registered with the International Platform of Registered Systematic Review and Meta-Analysis Protocols (INPLASY202270025, https://inplasy.com/inplasy-2022-7-0025/).

Literature search
As the search strategy shown in Supplementary Table 2, two independent reviewers (YXT and LLT) comprehensively searched eligible trials in the following electronic databases (PubMed, Medline, Embase, and The Cochrane Library) from inception to November 15, 2023, without language restrictions.Furthermore, additional potentially relevant publications were manually searched based on the reference lists of identified trials.Ongoing trials were also searched on the International Clinical Trials Registry Platform (ClinicalTrials.gov).

Inclusion and exclusion criteria
All studies were assessed independently by two reviewers (YXT and LLT).Inclusion criteria: (1) randomized controlled trials (RCTs), non-randomized clinical trials, prospective or retrospective cohort studies that investigated the impact of anesthetic agents on long-term survival during cancer surgery; (2) adult patients undergoing resection of the malignant tumor under general anesthesia; (3) studies that compared propofol-based anesthesia with inhalation anesthesia such as halothane, enflurane, isoflurane, sevoflurane or desflurane; (4) studies that reported overall survival (OS), which was defined as from the date of surgery to the date of death from any cause, or recurrence-free survival (RFS), which was defined as from the date of surgery to the date of first recurrence; (5) studies that reported hazard ratio (HR) and 95 % confidence interval (CI).Exclusion criteria: protocols, comments, reviews, case reports, casecontrol studies, and studies with insufficient data were excluded.Disagreements were resolved through discussion with a third reviewer (BC).

Study selection
The eligible studies were evaluated independently by two reviewers (YXT and LLT).Initially, the duplicate studies were eliminated by EndNote X9 software (Thomson Reuters, New York).Secondly, the remaining studies were screened through reading titles and abstracts.Finally, full texts were further reviewed for inclusion.Disagreements were reconciled through discussion with a third reviewer (BC).

Data collection
Two reviewers (YXT and YTR) independently extracted relevant data using a standard extraction data form.The following information was extracted from each eligible study: (1) general information included study design, first author's name, publication year, and country; (2) characteristics of participants included the number of patients, type of surgery, elective or emergency nature of surgery; (3) anesthetic protocol in each group; (4) outcomes included follow-up time, HR and 95 % CIs of reported outcomes.We would contact the original authors by email if relevant data were unavailable.Disagreements were reconciled through discussion with a third reviewer (BC).

Risk of bias in individual studies
Two reviewers (YXT and YTR) independently assessed the quality of the included studies.The randomized control trials (RCTs) were evaluated using the Cochrane risk of bias assessment tool.The evaluation contents included the following aspects: random Y. Tang et al. sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases.Each potential bias was graded as "low risk of bias," "high risk of bias," or "unclear risk of bias".On the other hand, the observational studies were evaluated based on the Newcastle Ottawa Scale (NOS), which includes three domains with a total score of 9 points: selection of subjects (4 points), comparability of the groups (2 points), as well as exposure and outcome ascertainment (3 points).Studies with a NOS score ≥7 were considered to have a lower risk of bias [26].In case of disagreements, a consensus was reached through discussion with a third reviewer (BC).

Statistical analysis
Review Manager 5.4 software (Cochrane, UK) and Stata 16 (StataCorp LLC, College Station, TX) were used to conduct this metaanalysis.We used the adjusted HRs when both unadjusted and adjusted HRs were provided.To maintain inhalation anesthesia as a reference group, we inverted the HRs and 95 % CIs where intravenous anesthesia was reported as a reference group.Heterogeneity was assessed through P and I 2 .Meta-analyses were performed using the random-effects model as we have identified clinical and methodological heterogeneity among studies.To explore the underlying origins of heterogeneity, subgroup analyses were performed based on different types of study centers, cancer surgery, inhalation anesthetics, and statistical models.Publication bias was evaluated using Funnel plots and Egger's test.P values < 0.05 were considered statistically significant.

Study selection
The initial search retrieved 2901 records.Two additional studies were identified from the reference lists.After removing duplicates, 1740 records remained.Of these, 1689 records were excluded through reading titles and abstracts.We further reviewed the full text of the remaining 51 records, the HR data about the outcome was unavailable in 6 studies [27][28][29][30][31][32], and 1 study had contradictory data on the outcome [33].Hence, 44 studies (42 retrospective cohort studies and 2 RCTs) met the eligibility criteria (details of the flow chart are shown in Fig. 1).Furthermore, we searched the clinical trial registry platforms and identified nine ongoing RCTs.

Risk of bias assessment
Risk of bias assessments of the retrospective studies are displayed in Table 2. Twelve studies received 7 stars, twenty-five studies received 8 stars, and five studies received 9 stars.The quality of all retrospective studies had a low risk of bias.Both the two multicenter RCTs described randomization and allocation concealment, presented complete data, and had no selective reporting.Due to the inherent nature of study intervention, it was not feasible to blind anesthesiologists to group assignment.Nevertheless, the blinding of participants, outcomes assessors, and other team members were reported in two RCTs.As a result, both included RCTs had a low risk of bias in each aspect and were considered high quality.

Publication bias
The asymmetric funnel plot showed that there was a high risk of publication bias both in OS (Fig. 2A) and RFS (Fig. 2B).Moreover, the Egger's test also indicated a significant publication bias for OS (P = 0.001) and RFS (P = 0.004).
Y. Tang et al.

Discussion
This study investigated whether propofol-based anesthesia or inhalation anesthesia might impact long-term oncological outcomes.We found that propofol-based anesthesia may be associated with improved OS and RFS compared with inhalation anesthesia in all cancer sites, which is partially consistent with the result of an important previous meta-analysis [16].Chang et al. reported that propofol-based anesthesia was associated with better OS but not RFS than volatile anesthesia in all cancer types [16].Moreover, growing in vivo and in vitro studies have provided possible explanations to support this claim.First, anesthetics may have different modulatory effects on immune function.Preclinical studies have demonstrated that propofol could protect anti-tumor immunity by preserving the activity and cytotoxicity of circulating NK cells [9,72].In addition, propofol has been shown to enhance the cytotoxic activity of T-lymphocytes [10], facilitate the activation and differentiation of T helper (Th) cells [73], and retain the Th1/Th2 ratio [74].By contrast, multiple studies have suggested that inhalation anesthetics may suppress the immune system by reducing NK-cell cytotoxicity [72], declining the Th1/Th2 ratio [6,74], and inducing apoptosis in T lymphocytes [75].Second, anesthetics may exert direct effects on cancer cells.Propofol has been demonstrated to inhibit the growth and invasion of various cancer cells.This is achieved through different pathways such as increasing the expression of miR-219-5p in hepatocellular carcinoma cells [11], upregulating miR-124-3p.1 and downregulating AKT3 in colorectal cancer [12], suppressing the transcription factor slug in ovarian cancer cells [76], or modulating ERK-VEGF/MMP-9 signaling in Eca-109 esophageal squamous cells [77].
In the subgroup analysis based on the types of study centers, it is worth noting that propofol-based anesthesia had better OS and RFS than inhalation anesthesia only in single-center studies but not in multicenter studies, which was not analyzed in the previous metaanalysis [16].To achieve a significant difference effect on OS and/or RFS in single-center studies with small sample sizes, an implausibly large effect size seems to be very important [78].Therefore, such limited external validity may potentially diminish the magnitude of the treatment effects observed in multicenter studies compared to single-center studies.A meta-epidemiological study and a systematic review have reported that single-center studies showed significantly larger treatment effects than multicenter studies [79,80].Interestingly, when several multicenter studies were combined, propofol-based anesthesia showed no advantage in OS and RFS, which is in keeping with the findings of two multicenter RCTs [70,71].One RCT [70] found that OS and RFS were comparable between propofol-based and sevoflurane-based anesthesia in mixed cancers.Another RCT [71] detected no significant difference in OS between propofol-based and sevoflurane-based anesthesia for breast cancer surgery.Our conflicting result based on the types of study centers may be attributable to the fact that multicenter studies have larger sample sizes and higher levels of collaboration and cross-validation, while single-center studies may have limited scope and potential biases.Therefore, more prospective multicenter RCTs are warranted to determine the optimal anesthetic choice for cancer surgery.On the other hand, in the subgroup analysis based on the types of cancer, we also found no benefit of propofol-based anesthesia for OS and RFS in mixed cancer and breast cancer surgeries, which is also consistent with the findings of two RCTs.Different cancers may have distinct basic characteristics such as proliferation, invasion, and metastatic abilities, leading to varying survival rates and recurrence.Considering the wide range of cancers and surgical procedures, more well-sized RCTs regarding different types of cancer are needed to further investigate the effects of general anesthetics on long-term survival in cancer surgery.
In addition, we also found differential survival effects based on inhalation anesthetics.Receiving propofol resulted in significantly improved OS and RFS compared to desflurane and mixed anesthetics.However, propofol-based anesthesia only had better OS but not RFS compared to sevoflurane-based anesthesia.Similarly, the impact of inhalation anesthetics on cancer cell biology remains conflicting.Isoflurane could promote the survival and migration of human glioblastoma stem cells and bladder cancer cells [81,82].Sevoflurane increased invasion and migration in glioblastoma, cervical, and prostate cancer cells [83][84][85], but not in colorectal cancer, ovarian cancer, colon cancer, breast cancer, and lung cancer cells [86][87][88][89][90]. Analogously, desflurane also showed conflicting results.Desflurane was shown to promote the migration of colorectal cancer and ovarian cancer cells [91,92], but not in glioma cells [93].These results indicate that the effects of different inhalation anesthetics on cancer cell biology are different.
Our meta-analysis had several limitations.First, the majority of included studies were retrospective observational studies.The retrospective design has an inherent weakness, which cannot be compensated by a meta-analysis of ever so many retrospective studies.Although most retrospective studies performed propensity score matching, there may have been other unpredicted confounding factors and selection bias, making it impossible to prove a causal association.Second, one major weakness of this study is the strong potential for publication bias, which could be attributed to the inclusion of many more single-center studies than multicenter studies.Furthermore, the same research group conducted eight single-center retrospective studies at different times, further increasing the risk of biases.Therefore, our results should be interpreted with caution while considering this limitation.Third, RFS is a somewhat tricky endpoint as most of the included studies did not clearly explain how to define recurrence.Moreover, to be valid, autopsies must be conducted at a frequency of 100 % or close to it, with a rigorous focus on recurrence.Due to the low frequency of autopsies, there may be some inherent uncertainty in the results of RFS.Fourth, cancer staging data is crucial for long-term oncological prognosis, but several included studies lacked this information [24,35,56,68].The unavailability of tumor staging characteristics can be a significant confounding factor.Finally, surgical techniques and clinical care for cancer patients have improved, resulting in better prognosis.However, some studies included patients over ten years, which may be an additional confounding factor [21,34,52,54,66].

Conclusions
In conclusion, our current systematic review and meta-analysis indicated that propofol-based anesthesia may have a favorable effect on overall survival and recurrence-free survival compared with inhalation anesthesia in some cancer surgeries.However, considering the inherent weaknesses of retrospective designs and the strong publication bias, caution should be taken when Y. Tang et al. interpreting the results.Well-designed multicenter randomized controlled trials regarding different types of cancer are still needed to clarify the association between general anesthetics and long-term oncological outcomes.Extraordinarily, future studies should be stratified according to different cancer sites, tumor stages, surgical types, and basic characteristics of the patient.

Fig. 1 .
Fig. 1.Flow chart of included and excluded studies.
Y.Tang et al.

Table 1
Characteristics of included studies.

Table 2
Risk of bias assessment based on the Newcastle-Ottawa scale.