Secondary Logo

Journal Logo

Review Articles: Meta-Analysis

Inflammatory Biomarker Levels After Propofol or Sevoflurane Anesthesia: A Meta-analysis

O’Bryan, Liam J. MD*; Atkins, Kelly J. DPsy*,†; Lipszyc, Adam MD*; Scott, David A. PhD*,†; Silbert, Brendan S. MB, BS*,†; Evered, Lis A. PhD*,†,‡

Author Information
doi: 10.1213/ANE.0000000000005671



  • Question: How does the choice of general anesthetic technique affect the perioperative inflammatory response?
  • Findings: Despite an overall rise in inflammatory biomarkers postoperatively, the choice of propofol or sevoflurane for maintenance of general anesthesia had no effect on perioperative biomarker levels.
  • Meaning: The contribution of patient and surgical factors far outweighs the influence of anesthetic agents on the perioperative inflammatory response.

Surgery is known to cause tissue damage, initiating both an inflammatory and immunomodulatory response.1 The immunomodulatory role of anesthesia in the perioperative inflammatory response is unclear, with in vitro data suggesting an immunosuppressive action.2,3 Anesthetic agents have been associated with impaired neutrophil and monocyte function, depression of lymphocyte proliferation, and variable findings in terms of inflammatory biomarker release.2 Early research into the surgical stress response focused on the endocrine axis, especially cortisol levels, finding significant elevation in keeping with the severity of surgery.4 Since the establishment of enzyme-linked immunosorbent assay (ELISA) techniques in 1971, the ability to quantify inflammatory cytokines has allowed researchers to identify contributors to the perioperative inflammatory response.5 Quantifying the inflammatory response to surgery is important; perioperative sequelae including neurocognitive dysfunction and recurrence of oncological disease may be mediated by inflammation.2,6 The agent used for maintenance of general anesthesia has been suggested to influence immune system–related outcomes such as micrometastases during cancer surgery, which affect cancer recurrence,7 and neuroinflammation contributing to perioperative neurocognitive disorders (PND).8

Perioperative inflammation may contribute to cognitive decline, particularly in vulnerable or elderly populations.1,8 A systematic review by Paredes et al9 determined the incidence of postoperative cognitive dysfunction (POCD) at 3 months postsurgery (excluding cardiac and intracranial surgery) to be 11.7%.9 In addition, more than 20% of older patients experience undiagnosed cognitive impairment and thus present for surgery at an increased risk of postoperative delirium (POD) and poor outcomes.10 Of patients affected by PND, many experience cognitive decline for up to 2 years following surgery.11 In the long-term, PND and POD are associated with increased risk of adverse outcomes including mortality, risk of dementia, premature loss of work, and functional dependence.8,12,13

Recent evidence illustrates a clinical correlation between the inflammatory response and PND; however, data regarding the relative contribution of anesthetic agents are lacking.14–17 Two systematic reviews comparing total intravenous anesthesia (TIVA) and inhaled anesthetic agents with respect to what was previously defined as POCD found insufficient evidence to determine a difference in effect between techniques.18,19

We conducted a systematic review and meta-analysis of randomized controlled trials with an aim to evaluate the effect of propofol and sevoflurane on perioperative inflammatory biomarker levels. Identifying any difference between agents is significant because it may provide insights into the physiological effects of the 2 most commonly used forms of general anesthesia that are not overtly apparent in the clinical setting. Our secondary objective was to identify the incidence of PND after propofol or sevoflurane anesthesia to correlate its associations with perioperative inflammation and with choice of anesthetic agent.


This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.20 The PRISMA checklist can be found in Supplemental Digital Content 1, Table 1, The review methodology was predefined and registered with the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42020198351).

Study Characteristics


Included studies were of patients undergoing general anesthesia for a surgical procedure. Studies of patients undergoing intracranial neurosurgery were excluded due to possible confounders impacting cognitive assessment.


Our intervention of interest was a comparison between TIVA using propofol and inhalational anesthesia using sevoflurane. Anesthetic agent was determined according to that used for maintenance of anesthesia, not induction. Studies of mixed general anesthesia (in which propofol was given alongside volatile anesthesia throughout surgery) were excluded.


Included studies reported the perioperative levels of inflammatory cytokines in patient sera. The inflammatory biomarkers chosen as outcomes were C-reactive protein (CRP) and cytokines interleukin (IL)-6, IL-10, and tissue necrosis factor alpha (TNF-α). We chose these markers based on their high frequency in both in vitro and in vivo perioperative studies. Studies that measured only intraoperative inflammatory biomarker levels21 or biomarkers in tissues other than blood (eg, airway epithelium)22,23 were excluded. Time points for cytokine measurement were grouped into T0 (preoperative baseline), T1 (0–2 hours postoperatively), T2 (12–24 hours postoperatively), and T3 (2–4 days postoperatively). We chose these time points to align with the outcomes most commonly used in identified relevant studies. To limit heterogeneity across sampling periods, we excluded studies that measured inflammatory biomarkers outside of our specified timepoints. The secondary outcome of interest was any clinical measure of PND or POD, assessed using cognitive screening tools such as the mini-mental state examination (MMSE) or full neuropsychological test batteries.

Study Design

Only randomized controlled trials were considered for inclusion in the systematic review. No other limitations were placed on study design. Studies in languages other than English were translated using the Google Translate function. Where translation was not adequate for data extraction, these studies were excluded.

Search Strategy

MEDLINE and Embase were searched for relevant published literature. The search was performed on December 7, 2020 using the Ovid platform. Medical Subject Heading (MeSH) terms were used alongside free text. Snowballing was performed to identify additional relevant references from study bibliographies. Supplemental Digital Content 1, Tables 2A and 2B,, include full descriptions of the search strategies for each database.

Study Selection

Studies were screened for eligibility by 3 independent reviewers (L.J.O., K.J.A., A.L.). Studies that met inclusion criteria based on title and abstract, or of which inclusion was uncertain, were moved to full-text assessment. Each full text was assessed in duplicate to determine eligibility for inclusion. Reviewers were not blinded to study authors or institutions. When 2 reviewers disagreed on study eligibility, a third reviewer was consulted to make the final decision on inclusion.

Data Management

Covidence software (Veritas Health Innovation Ltd) was used to identify duplicates and to aid study selection and data extraction. Data extracted included authors, funding, clinical setting, patient population, surgical type, duration of anesthesia, perioperative cytokine levels, and cognitive assessment (Table). For risk-of-bias assessment, the revised Cochrane Risk of Bias 2 (ROB2) tool through the ROBVis interface24 was used to categorize each domain as low, unclear, or high risk of bias.25 Data extraction was performed in duplicate, and extraction tables were compared to develop a consensus table. If differences in data extraction were found, the data were rereviewed to identify and correct sources of discrepancy.

Table. - Population and Methods of Included Trials
Author (year) N population Surgery type Duration of anesthesia: propofol versus sevoflurane (min) Biomarkers assessed Time points Cognitive tests
Abramo et al (2012)33 30 obese adult patients Roux en Y gastric bypass 154 (134–200) vs 150 (120–190) IL-6, IL-10, TNF-α T0, T1, T2 Nil
Ammar and Mahmoud (2016)34 50 adult patients Elective AAA repair 180 ± 75 vs 184 ± 81
Chen et al (2016)35 50 pediatric patients age 2–12 Flexible bronchoscopy 19 ± 7 vs 17 ± 5 IL-6, IL-10, and TNF-α T0, T1, T2, T3 Nil
Conzen et al (2003)36 20 adult patients Elective off pump CABG 351 ± 90 vs 334 ± 93 IL-6, CRP T0, T1, T2 Nil
De La Gala et al (2017)37 180 adult patients Lung resection surgery 290 (231–353) vs 302 (220–365) IL-6, IL-10, and TNF-α T0, T2 Nil
El Azab et al (2002)38 30 adult patients Elective CABG 230 ± 60 vs 205 ± 45 IL-6 and TNF-α T0, T1, T2 Nil
Geng et al (2017)39 150 adult patients aged ≥65 Laparoscopic cholecystectomy 73 (64–109) vs 88 (70–113) IL-6 and TNF-α T0, T1, T2 Test battery
Ihn et al (2009)40 65 female adult patients Elective total abdominal hysterectomy 161 ± 46 vs 143 ± 23 IL-6 T0, T1 Nil
Kvarnström et al (2012)41 50 adult patients Elective open colorectal surgery 292 (239–354) vs 333 (265–338) IL-6, IL-10, and TNF-α T0, T1, T2 Nil
Lee et al (2012)42 48 patients aged 40–70 Elective Ivor Lewis operation 313 ± 41 vs 331 ± 40 IL-6 and CRP T0, T1, T2 Nil
Lim et al (2018)43 44 female patients aged >20 Breast cancer surgery 132 (109–155) vs 128 (115–196) IL-6, IL-10, and TNF-α T0, T1, T2 Nil
Lindholm et al (2015)44 193 adult patients Elective open abdominal aortic surgery 227 ± 71 vs 211 ± 52 IL-6 and CRP T0, T1, T2 Nil
Matsota et al (2018)45 34 patients aged 20–65 Inguinal hernia repair or varicose vein stripping 122 ± 50 vs 108 ± 36 IL-6 and TNF-α T0, T1 Nil
Micha et al (2016)46 80 patients aged 60–74 Noncardiologic, non-neurosurgical tumor resection 141 (130–149) vs 148 (125–153) IL-6, IL-10, TNF-α, and CRP T0, T3 MMSE, CAM, BDI
Potočnik et al (2014)47 40 patients aged 20–70 Thoracic surgery with OLV 142 ± 47 vs 139 ± 55 IL-6, IL-10, and CRP T0, T1, T2 Nil
Qiao et al (2015)48 90 patients aged 65–75 Elective resection of esophageal carcinoma 130 ± 5 vs 128 ± 8 IL-6 and TNF-α T0, T2, T3 MMSE
Sahoo et al (2019)49 66 adult patients Spinal surgery 177 ± 36 vs 181 ± 27 IL-6 and TNF-α T0, T3 Test battery
Schilling et al (2011)50 63 adult patients Thoracic surgery 122 (48–252) vs 109 (39–196) IL-6, IL-10, and TNF-α T0, T1 Nil
Schneemilch et al (2005)51 50 adult patients Partial discectomy 137 (80–235) vs 139 (75–225) IL-6 T0, T1, T3 Nil
Tian et al (2017)52 62 adult patients Lung cancer resection NA IL-6, IL-10 T0, T2 MMSE
Wakabayashi et al (2014)53 28 patients aged 52–78 Esophagectomy with OLV 530 ± 78 vs 485 ± 61 IL-6, T0, T2 Nil
Yang et al (2017)54 76 adult patients Valve replacement surgery 198 ± 28 vs 183 ± 34 IL-6, IL-10 T0, T2, T3 Nil
Yoo et al (2014)55 112 adult patients Valvular heart surgery 188 ± 46 vs 187 ± 49 IL-6 T0, T1, T2 Nil
Abbreviations: AAA, abdominal aortic aneurysm; BDI, Beck Depression Inventory; CABG, coronary artery bypass graft; CAM, Confusion Assessment Method; CRP, C-reactive protein; IL, interleukin; MMSE, mini-mental state examination; OLV, 1-lung ventilation; T0, baseline preoperative; T1, immediate postoperative (0–2 h); T2, day 1 postoperative (12–24 h); T3, delayed postoperative (48–96 h); TNF-α, tissue necrosis factor alpha.

Figure 1.:
Inflammatory biomarker levels across the perioperative period reported as mean ± standard error for (A) IL-6, (B) TNF-α, (C) CRP, and (D) IL-10. CRP indicates C-reactive protein; IL, interleukin; TNF-α, tissue necrosis factor alpha.

Study results were extracted as mean and standard deviation (SD) for each time point reported, and authors were contacted for additional data if these were not available. If additional data were not available, study figures were analyzed using Plot Digitiser open-source software to obtain mean and SD. This method has been correlated with a high degree of accuracy and efficiency when compared with manual extraction.26 When median (interquartile range [IQR]) were reported, mean and SD were calculated using the methods described by Wan et al,27 which have been reported to minimize bias in the presence of skewed data. For data pertaining to PND, either the mean and SD of specific test outcomes or incidence of PND as reported by each study were extracted.

Statistical Analysis

Data were analyzed using the metafor28 and dmetar29 packages of R core V3.6.130. Separate meta-analyses were performed of studies examining IL-6, IL-10, TNF-α, and CRP cytokine responses at T0, T1, T2, and T3. We used Hartung-Knapp-Sidik-Jonkman random effects models with a Hartung-Knapp adjustment to account for anticipated heterogeneity between surgical populations.31 For each meta-analysis, the mean difference with 95% confidence interval (95% CI) of cytokine levels between anesthetic groups is reported. An α level of .05 was set for all significance tests. If fewer than 3 studies reported a specific outcome and time point, these data were not pooled. Heterogeneity was calculated using the I2 statistic, for which a result ≧75% suggests significant statistical heterogeneity between studies.32 To evaluate sources of clinical heterogeneity, we performed Hartung-Knapp adjusted metaregressions when the number of pooled studies was ≥10. Surgery type, surgery duration, patient age, and their interactions were entered as covariates in all models. All metaregression models underwent permutation tests. To visually depict the time course of the overall inflammatory response to surgery across studies, the mean value for each biomarker at each time point was presented graphically (Figure 1). For outcomes pertaining to clinical measures of PND, the data are presented descriptively.


In the initial search, 1152 studies were identified. Following screening of abstracts and full texts by the review team, a total of 23 studies remained and were included in the review (Supplemental Digital Content 2, Figure 1,

Characteristics of included studies are presented in the Table. Included studies assessed a range of patient populations and surgical types, including 4 studies of patients undergoing cardiac surgery36,38,54,55 and 8 of patients with malignancy.37,41–43,46–48,50,52,53 Median duration of surgery ranged significantly from under 30 minutes in 1 study of flexible bronchoscopy in children35 to over 300 minutes in patients undergoing esophagectomy.42,53 The age of patients also differed significantly with 1 study specific to a pediatric population,35 and 13 with a median age of more than 60 years.34,36–39,42,44,46,48,50,53,56,57 Within all studies, baseline characteristics including patient age, gender, duration of anesthesia, and weight or body mass index (BMI) were similar between groups receiving propofol and sevoflurane. A range of inflammatory biomarkers was assessed across all studies; 5 studies assessed cognition, and 2 of these used only the MMSE for assessment.

The risk-of-bias assessment is shown in Supplemental Digital Content 3, Figure 2, Overall, risk of bias was low across studies. The most common concerns for risk of bias were inadequate description of participant allocation concealment. Given the objective nature of assay analysis, we determined that failure to blind participants, personnel, or outcome assessors did not introduce any significant risk of bias for the assessment of inflammatory cytokines.

Primary Outcome: Inflammatory Biomarker Response

Twenty-one studies reported IL-6 levels at one or more of our included timepoints, 10 studies reported IL-10, 12 studies TNF-α, and 5 studies CRP. Requests for additional data were made of 15 corresponding authors33,38,40–47,50–53,55; we received complete data from 2 of these authors.46,50 Data were converted from median with IQR to mean and SD in 7 studies33,41–44,47,51 using the method described by Wan et al.27 Plot Digitiser was used to interpret figures for 9 studies.38,40,42,44,45,51–53,55

Figure 1 demonstrates the trend of inflammatory biomarkers perioperatively; this was established by taking the mean of biomarker levels reported at each time point and does not represent a comparison of propofol and sevoflurane over time. The average of biomarker levels reported at each time point shows an overall rise in IL-6 and TNF-α in the immediate postoperative period (T1), which began to normalize over the later time points and fell below baseline levels after 2 to 4 days (T3; Figure 1A, B). Average IL-10 levels showed a slight and gradual increase between baseline (T0) and day 1 postoperatively (T2; Figure 1D). Average CRP levels showed a delayed peak at T3 (Figure 1C). It is important to note that data for T3 were taken from a single study only. These findings are consistent with those previous studies of perioperative inflammation, which demonstrated a rise in inflammatory biomarkers before the rise in CRP.58,59

Levels of IL-6 did not differ between patients receiving propofol compared to sevoflurane in the immediate perioperative period (T1), meandiff = 5.36 (−16.85, 27.58), P = .62, I2 = 91%; at day 1 (T2), meandiff = −11.07 (−35.96, 13.83), P = .36, I2 = 97%; and after 48 to 96 hours (T3), meandiff = −3.43 (−10.54, 3.68), P = .27, I2 = 94.5% (Figure 2). IL-10 levels showed no statistical difference between anesthetic groups at T1, meandiff = 1.61 (−7.03, 10.26), P = .68, I2 = 98.5% or at T2, meandiff = 0.51 (−3.68, 4.69), P = .78, I2 = 97.8% (Figure 3). Likewise, TNF-α did not differ between groups at any time point (T1: meandiff = −2.51 [−8.46, 3.45], P = .37, I2 = 90%; T2: meandiff = −18.61 [−59.35, 22.13], P = .32, I2 = 94%; T3: meandiff = −5.30 [−12.95, 2.35], P = .12, I2 = 94%) nor did levels of CRP (T2: meandiff=1.87 [−16.39, 20.14], P = .77, I2 = 95%; Figures 4 and 5). As only 2 studies assessed CRP at T1, their findings were not pooled.36,44 Both found no significant difference between anesthetic technique, meandiff = 0.30 (−0.01, 0.60)36 and meandiff = −0.19 (−1.00, 0.62).44 Similarly, 2 studies assessed IL-10 levels at T3, finding no difference between sevoflurane and propofol.46,54

Figure 2.:
Forest plot of postoperative IL-6 levels at (A) immediate postoperative (0–2 h), (B) day 1 postoperative (12–24 h), and (C) delayed postoperative (48–72 h). CI indicates confidence interval; IL, interleukin; SD, standard deviation.
Figure 3.:
Forest plot of postoperative IL-10 levels at (A) immediate postoperative (0–2 h) and (B) day 1 postoperative (12–24 h). CI indicates confidence interval; IL, interleukin; SD, standard deviation.
Figure 4.:
Forest plot of postoperative TNF-α levels at (A) immediate postoperative (0–2 h), (B) day 1 postoperative (12–24 h), and (C) delayed postoperative (48–72 h). CI indicates confidence interval; SD, standard deviation; TNF-α, tissue necrosis factor alpha.
Figure 5.:
Forest plot of postoperative CRP levels at day 1 postoperative (12–24 h). CI indicates confidence interval; CRP, C-reactive protein; SD, standard deviation.

In all meta-analyses performed, there was a high degree of heterogeneity as indicated by the I2. To understand the extent to which patient age, surgical procedure, duration of surgery or their interactions contributed to heterogeneity, we performed metaregressions, with permutation tests for each biomarker at each time point. Metaregression results for IL-6 are shown in Supplemental Digital Content 1, Table 3, When evaluating IL-6 at T1, F(6, 10) = 5.13, P = .01, these factors accounted for 68.36% of the heterogeneity and for 58.80% at T2. After permutation tests, our model did not significantly explain the heterogeneity in other biomarkers or time points, which may be due to differences in a range of unmeasured variables in patient populations. Despite this degree of heterogeneity, study results were combined in a random effects meta-analysis, as they reported homogenous outcomes and time points, albeit in a range of patient populations.

Secondary Outcome: PND

A total of 5 studies assessed perioperative changes in cognitive function.39,46,48,49,56 Two studies evaluated cognition using only the MMSE, and the remaining 3 evaluated cognitive outcomes with varying neuropsychological test batteries. A meta-analysis of cognitive outcomes was not performed due to the paucity of studies reporting homogenous outcomes. However, 4 of the 5 studies reported a significant difference in postoperative neurocognitive function favoring the use of propofol.39,46,48,56 The extent that PND correlated with inflammatory biomarkers was inconsistent across studies.


Taking the mean of reported levels revealed an increase in average inflammatory biomarker levels following surgery, with an early peak in the cytokines IL-6, IL-10, and TNF-α followed by a delayed peak in CRP. Our meta-analysis found no difference between propofol or sevoflurane regarding their effect on inflammatory biomarker levels. This absence of effect was observed despite pooling data for more than 1600 randomized participants, with overall low risk of bias across studies. We conducted a number of secondary analyses to test for an effect of patient and surgical variables on our outcomes of interest. Our analyses found that when studies were grouped based on participant age, surgical type (oncological, cardiac, and elective) or duration of anesthesia, there remained no difference between propofol or sevoflurane. Between included studies, there was a large degree of heterogeneity in results. High heterogeneity has been reported in previous systematic reviews of the perioperative inflammatory response.60 This heterogeneity suggests that the impact of patient and surgical variables, both measured and unmeasured, on the biomarker response likely far outweighs the impact of anesthetic agents.

TIVA using propofol is an increasingly utilized anesthetic technique, in part due to its possible benefits for patients with cancer and its favorable environmental impact compared to volatile anesthetics.7,61 There is significant interest in the potential for TIVA to be safer than volatile anesthesia for patients at risk of PND; however, this has yet to be demonstrated.18 To date, a number of randomized trials reporting the association PND and POD with TIVA and volatile agents have exhibited contradictory results.62,63 Existing studies used a range of outcomes, time points, and populations, making it a challenging area for meta-analysis. As a result, assessment of surrogate outcomes such as perioperative inflammation may provide valuable information regarding the impact of anesthetics on cognitive health and help focus future research. Despite a good number of high-quality randomized trials assessing the inflammatory response to surgery, only 1 other meta-analysis has evaluated the effect of anesthetic agents on the systemic inflammatory response.64 The present study was not specific in comparing TIVA and inhalational anesthesia and found insufficient data to adequately assess the impact of propofol and sevoflurane on the perioperative inflammatory response.

In vivo murine studies have previously demonstrated the link between postoperative inflammation and impaired cognitive function, with IL-6 and TNF-α specifically implicated.65,66 One randomized controlled trial that compared general anesthesia to regional anesthesia (without sedation or opioids) in 100 patients undergoing extracorporeal shockwave lithotripsy found no difference in the incidence of POCD at 3 months postoperatively but did not measure inflammatory biomarkers.67 Another recent trial compared pre- and postanesthetic cytokine levels in patients exposed to 2 hours of inhalational anesthesia in the absence of surgery.68 The authors reported that anesthesia alone had no effect on cytokine response. These findings, together with the results of our meta-analysis, suggest that the role of anesthesia in both the perioperative inflammatory response and development of PND is small. Targeted interventions to optimize outcomes among vulnerable patients by minimizing surgical duration and perioperative stressors may be more important considerations in reducing the incidence of inflammatory-mediated sequelae. Nonetheless, our results did show a clear inflammatory response to surgery and anesthesia (using both propofol and sevoflurane). While the effect of general anesthetic technique had no bearing on the perioperative inflammatory response, interventions to minimize perioperative inflammation are a field rich for further studies.

In assessing the impact of anesthetic agents on clinical tests of cognition, we were unable to pool the results of studies due to a paucity of high-quality evidence. Only 5 studies reported results of both perioperative inflammatory biomarkers and clinical assessments of cognition.39,46,48,49,56 In these studies, a range of assessment tools and outcomes of interest was used. Two studies used only the MMSE as a means of assessing cognition. MMSE was designed as a screening test only and is an inadequate diagnostic tool for PND. Despite a paucity of data, all but 1 study assessing cognition as an outcome of interest favored propofol as an intervention. Interestingly, the study by Sahoo et al,49 which reported no difference between agents, was conducted on a patient population with a mean patient age between 37 and 39 years, whereas the other 4 studies were conducted in a far older population (age >65 years).

A recent systematic review noted that in 274 existing studies of POCD, diagnosis was based on 259 different cognitive assessment tools.69 Reviewed studies also varied in terms of follow-up time and diagnostic criteria. These sources of heterogeneity limit the interpretation of existing data surrounding PND, making them unsuitable for meta-analysis. PND is associated with long-term sequelae including ongoing impaired cognition, increased risk of dementia, increased mortality, and premature retirement from work.13 Trials of interventions to mitigate these sequelae may therefore provide long-term clinical and economic benefit. Given the substantial heterogeneity between study methodology and results in our review, further trials are needed to assess for PND using existing validated diagnostic tools and in a high-risk surgical patient population (adults >65 years of age). Elucidating any potential relationship between clinical outcomes and the observed rise in biomarker levels during the perioperative period was beyond the scope of this systematic review. The association between biomarkers and cognitive outcomes should be evaluated in future studies across a longer postoperative course.


Our meta-analysis combined data from randomized controlled trials, embracing a range of surgical procedures. We report our analytic limitations using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework70 to assess the certainty of our findings (Supplemental Digital Content 1, Appendix 4, Our meta-analysis combined data from randomized controlled trials, with no substantial risks of bias identified in included studies. In keeping with previous meta-analyses reporting inflammatory biomarker levels in different surgical populations, we came across substantial heterogeneity between the results of included studies, which may reflect variable inflammatory responses to different surgical procedures and durations. While these studies were comparable based on their design and outcomes, heterogeneity affects the validity of pooled results; therefore, the certainty of our findings was downgraded for inconsistency. Heterogeneity for IL-6 was largely accounted for by metaregressions using a number of clinical characteristics, and it was therefore not downgraded. There were insufficient numbers of procedures in specific surgical categories to enable differentiation between procedure types. Our primary outcome of interest was a difference in inflammatory biomarker response in a general surgical population. A number of included studies reported median rather than a mean for our primary outcome of interest, which may suggest a skew in the original study data. Our method of conversion has been shown to minimize bias; however, bias cannot be entirely eliminated. A sensitivity analysis with converted study results excluded found no effect on our results. Included studies reported homogenous outcomes in a range of patients and surgery types, and our primary outcomes were therefore not downgraded for indirectness. In grading the precision of our findings, outcomes and timepoints varied in terms of the consistency of results and range of their CIs. An assessment for publication bias using Egger’s test found no significant publication bias for any outcome or time point.


Our findings suggest that the choice of maintenance anesthetic agent has little impact on the perioperative inflammatory response, as measured by the levels of IL-6, IL-10, TNF-α, and CRP. In contrast, the effects of individual patient and procedural factors may play a larger role in the inflammatory response. Thus, in the clinical setting, the optimization of patient and procedural variables is likely to have a more substantial impact on the perioperative inflammatory response than the choice of anesthetic agent. With regard to PND, the evidence base is less clear, and high-quality trials using validated diagnostic tools are needed to determine the difference, if any, between TIVA and inhalational anesthesia in terms of perioperative cognition.


The authors acknowledge the assistance of Dr Roman Kluger of the Department of Anesthesia for his expertise in planning our statistical analyses.


Name: Liam J. O’Bryan, MD.

Contribution: This author helped with study concept, study design, screening of papers, data extraction, drafting of manuscript, figures preparation, and final review.

Name: Kelly J. Atkins, DPsy.

Contribution: This author helped with screening of papers, data extraction, statistical analysis, drafting of manuscript, figures preparation, and final review.

Name: Adam Lipszyc, MD.

Contribution: This author helped with screening of papers, data extraction, drafting of manuscript, and final review.

Name: David A. Scott, PhD.

Contribution: This author helped with study concept and design, drafting of manuscript, and final review.

Name: Brendan S. Silbert, MB, BS.

Contribution: This author helped with study concept and design, drafting of manuscript, and final review.

Name: Lis A. Evered, PhD.

Contribution: This author helped with study concept and design, drafting of manuscript, and final review.

This manuscript was handled by: Robert Whittington, MD.


    1. Lord JM, Midwinter MJ, Chen YF, et al. The systemic immune response to trauma: an overview of pathophysiology and treatment. Lancet. 2014;384:1455–1465.
    2. Schneemilch CE, Schilling T, Bank U. Effects of general anaesthesia on inflammation. Best Pract Res Clin Anaesthesiol. 2004;18:493–507.
    3. McBride WT, Armstrong MA, McBride SJ. Immunomodulation: an important concept in modern anaesthesia. Anaesthesia. 1996;51:465–473.
    4. Prete A, Yan Q, Al-Tarrah K, et al. The cortisol stress response induced by surgery: a systematic review and meta-analysis. Clin Endocrinol (Oxf). 2018;89:554–567.
    5. Aydin S. A short history, principles, and types of ELISA, and our laboratory experience with peptide/protein analyses using ELISA. Peptides. 2015;72:4–15.
    6. Wall T, Sherwin A, Ma D, Buggy DJ. Influence of perioperative anaesthetic and analgesic interventions on oncological outcomes: a narrative review. Br J Anaesth. 2019;123:135–150.
    7. Yap A, Lopez-Olivo MA, Dubowitz J, Hiller J, Riedel B; Global Onco-Anesthesia Research Collaboration Group. Anesthetic technique and cancer outcomes: a meta-analysis of total intravenous versus volatile anesthesia. Can J Anaesth. 2019;66:546–561.
    8. Monk TG, Weldon BC, Garvan CW, et al. Predictors of cognitive dysfunction after major noncardiac surgery. Anesthesiology. 2008;108:18–30.
    9. Paredes S, Cortínez L, Contreras V, Silbert B. Post-operative cognitive dysfunction at 3 months in adults after non-cardiac surgery: a qualitative systematic review. Acta Anaesthesiol Scand. 2016;60:1043–1058.
    10. Scott DA, Evered L, Maruff P, MacIsaac A, Maher S, Silbert BS. Cognitive function before and after left heart catheterization. J Am Heart Assoc. 2018;7:e008004.
    11. Abildstrom H, Rasmussen LS, Rentowl P, et al. Cognitive dysfunction 1-2 years after non-cardiac surgery in the elderly. ISPOCD group. International study of post-operative cognitive dysfunction. Acta Anaesthesiol Scand. 2000;44:1246–1251.
    12. Inouye SK, Marcantonio ER, Kosar CM, et al. The short-term and long-term relationship between delirium and cognitive trajectory in older surgical patients. Alzheimer’s Dement. 2016;12:766–775.
    13. Steinmetz J, Christensen KB, Lund T, Lohse N, Rasmussen LS; ISPOCD Group. Long-term consequences of postoperative cognitive dysfunction. Anesthesiology. 2009;110:548–555.
    14. Tang JX, Baranov D, Hammond M, Shaw LM, Eckenhoff MF, Eckenhoff RG. Human Alzheimer and inflammation biomarkers after anesthesia and surgery. Anesthesiology. 2011;115:727–732.
    15. Kline R, Wong E, Haile M, et al. Peri-operative inflammatory cytokines in plasma of the elderly correlate in prospective study with postoperative changes in cognitive test scores. Int J Anesthesiol Res. 2016;4:313–321.
    16. Liu X, Yu Y, Zhu S. Inflammatory markers in postoperative delirium (POD) and cognitive dysfunction (POCD): a meta-analysis of observational studies. PLoS One. 2018;13:e0195659.
    17. Sieber FE, Neufeld KJ, Gottschalk A, et al. Effect of depth of sedation in older patients undergoing hip fracture repair on postoperative delirium: the STRIDE Randomized Clinical Trial. JAMA Surg. 2018;153:987–995.
    18. Miller D, Lewis SR, Pritchard MW, et al. Intravenous versus inhalational maintenance of anaesthesia for postoperative cognitive outcomes in elderly people undergoing non-cardiac surgery. Cochrane Database Syst Rev. 2018;8:CD012317.
    19. Sun H, Zhang G, Ai B, et al. A systematic review: comparative analysis of the effects of propofol and sevoflurane on postoperative cognitive function in elderly patients with lung cancer. BMC Cancer. 2019;19:1248.
    20. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol. 2009;62:e1–e34.
    21. Kawamura T, Kadosaki M, Nara N, et al. Effects of sevoflurane on cytokine balance in patients undergoing coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth. 2006;20:503–508.
    22. Kumakura S, Yamaguchi K, Sugasawa Y, et al. Effects of nitrous oxide on the production of cytokines and chemokines by the airway epithelium during anesthesia with sevoflurane and propofol. Mol Med Rep. 2013;8:1643–1648.
    23. Sugasawa Y, Yamaguchi K, Kumakura S, et al. Effects of sevoflurane and propofol on pulmonary inflammatory responses during lung resection. J Anesth. 2012;26:62–69.
    24. McGuinness LA, Higgins JPT. Risk-of-bias VISualization (robvis): an R package and Shiny web app for visualizing risk-of-bias assessments. Res Synth Methods. 2021;12:55–61.
    25. Higgins J, Savović J, Page MJ, Sterne JAC. RoB 2: a revised Cochrane risk-of-bias tool for randomized trials. Br Med J. 2019;366:l4898.
    26. Jelicic Kadic A, Vucic K, Dosenovic S, Sapunar D, Puljak L. Extracting data from figures with software was faster, with higher interrater reliability than manual extraction. J Clin Epidemiol. 2016;74:119–123.
    27. Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol. 2014;14:135.
    28. Viechtbauer W. Conducting meta-analyses in R with the metafor. J Stat Softw. 2010;36:1–48.
    29. Harrer M, Cuijpers P, Furukawa T, Ebert DD. dmetar: Companion R Package For The Guide’Doing Meta-Analysis in R’. R Packag version 00. 2019;9000.
    30. Team R Development Core. A Language and Environment for Statistical Computing. R Found Stat Comput. 2018;2. Accessed February 15, 2021.
    31. Röver C, Knapp G, Friede T. Hartung-Knapp-Sidik-Jonkman approach and its modification for random-effects meta-analysis with few studies. BMC Med Res Methodol. 2015;15:99.
    32. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA, eds. Cochrane Handbook for Systematic Reviews of Interventions. Version 6.2. Cochrane, 2021. Updated February 2021.
    33. Abramo A, Di Salvo C, Baldi G, et al. Xenon anesthesia reduces TNFα and IL10 in bariatric patients. Obes Surg. 2012;22:208–212.
    34. Ammar AS, Mahmoud KM. Comparative effect of propofol versus sevoflurane on renal ischemia/reperfusion injury after elective open abdominal aortic aneurysm repair. Saudi J Anaesth. 2016;10:301–307.
    35. Chen L, Cheng J, Wang YL. Effects of different anaesthetics on cytokine levels in children with community-acquired pneumonia undergoing flexible fibreoptic bronchoscopy. J Int Med Res. 2016;44:462–471.
    36. Conzen PF, Fischer S, Detter C, Peter K. Sevoflurane provides greater protection of the myocardium than propofol in patients undergoing off-pump coronary artery bypass surgery. Anesthesiology. 2003;99:826–833.
    37. De La Gala F, Piñeiro P, Reyes A, et al. Postoperative pulmonary complications, pulmonary and systemic inflammatory responses after lung resection surgery with prolonged one-lung ventilation. Randomized controlled trial comparing intravenous and inhalational anaesthesia. Br J Anaesth. 2017;119:655–663.
    38. El Azab SR, Rosseel PM, De Lange JJ, van Wijk EM, van Strik R, Scheffer GJ. Effect of VIMA with sevoflurane versus TIVA with propofol or midazolam-sufentanil on the cytokine response during CABG surgery. Eur J Anaesthesiol. 2002;19:276–282.
    39. Geng YJ, Wu QH, Zhang RW. Effect of propofol, sevoflurane, and isoflurane on postoperative cognitive dysfunction following laparoscopic cholecystectomy in elderly patients: a randomized controlled trial. J Clin Anesth. 2017;38:165–171.
    40. Ihn CH, Joo JD, Choi JW, et al. Comparison of stress hormone response, interleukin-6 and anaesthetic characteristics of two anaesthetic techniques: volatile induction and maintenance of anaesthesia using sevoflurane versus total intravenous anaesthesia using propofol and remifentanil. J Int Med Res. 2009;37:1760–1771.
    41. Kvarnström AL, Sarbinowski RT, Bengtson JP, Jacobsson LM, Bengtsson AL. Complement activation and interleukin response in major abdominal surgery. Scand J Immunol. 2012;75:510–516.
    42. Lee JJ, Kim GH, Kim JA, et al. Comparison of pulmonary morbidity using sevoflurane or propofol-remifentanil anesthesia in an Ivor Lewis operation. J Cardiothorac Vasc Anesth. 2012;26:857–862.
    43. Lim JA, Oh CS, Yoon TG, et al. The effect of propofol and sevoflurane on cancer cell, natural killer cell, and cytotoxic T lymphocyte function in patients undergoing breast cancer surgery: an in vitro analysis. BMC Cancer. 2018;18:159.
    44. Lindholm EE, Aune E, Seljeflot I, Otterstad JE, Kirkebøen KA. Biomarkers of inflammation in major vascular surgery: a prospective randomised trial. Acta Anaesthesiol Scand. 2015;59:773–787.
    45. Matsota P, Kostopanagiotou G, Kalimeris K, et al. Transient effects of anesthesia on leukocyte apoptosis and monocyte cytokine stimulation: a clinical study. Immunol Invest. 2018;47:327–334.
    46. Micha G, Tzimas P, Zalonis I, Kotsis K, Papdopoulos G, Arnaoutoglou E. Propofol vs sevoflurane anaesthesia on postoperative cognitive dysfunction in the elderly. A randomized controlled trial. Acta Anaesthesiol Belg. 2016;67:129–137.
    47. Potočnik I, Novak Janković V, Šostarič M, et al. Antiinflammatory effect of sevoflurane in open lung surgery with one-lung ventilation. Croat Med J. 2014;55:628–637.
    48. Qiao Y, Feng H, Zhao T, Yan H, Zhang H, Zhao X. Postoperative cognitive dysfunction after inhalational anesthesia in elderly patients undergoing major surgery: the influence of anesthetic technique, cerebral injury and systemic inflammation. BMC Anesthesiol. 2015;15:154.
    49. Sahoo AK, Panda N, Sabharwal P, et al. Effect of anesthetic agents on cognitive function and peripheral inflammatory biomarkers in young patients undergoing surgery for spine disorders. Asian J Neurosurg. 2019;14:1095–1105.
    50. Schilling T, Kozian A, Senturk M, et al. Effects of volatile and intravenous anesthesia on the alveolar and systemic inflammatory response in thoracic surgical patients. Anesthesiology. 2011;115:65–74.
    51. Schneemilch CE, Ittenson A, Ansorge S, Hachenberg T, Bank U. Effect of 2 anesthetic techniques on the postoperative proinflammatory and anti-inflammatory cytokine response and cellular immune function to minor surgery. J Clin Anesth. 2005;17:517–527.
    52. Tian HT, Duan XH, Yang YF, Wang Y, Bai QL, Zhang X. Effects of propofol or sevoflurane anesthesia on the perioperative inflammatory response, pulmonary function and cognitive function in patients receiving lung cancer resection. Eur Rev Med Pharmacol Sci. 2017;21:5515–5522.
    53. Wakabayashi S, Yamaguchi K, Kumakura S, et al. Effects of anesthesia with sevoflurane and propofol on the cytokine/chemokine production at the airway epithelium during esophagectomy. Int J Mol Med. 2014;34:137–144.
    54. Yang XL, Wang D, Zhang GY, Guo XL. Comparison of the myocardial protective effect of sevoflurane versus propofol in patients undergoing heart valve replacement surgery with cardiopulmonary bypass. BMC Anesthesiol. 2017;17:37.
    55. Yoo YC, Shim JK, Song Y, Yang SY, Kwak YL. Anesthetics influence the incidence of acute kidney injury following valvular heart surgery. Kidney Int. 2014;86:414–422.
    56. Wang G, Liu J, Gao J, Zheng X. Comparison of the effects of sevoflurane and propofol anesthesia on pulmonary function, MMP-9 and postoperative cognition in patients receiving lung cancer resection. Oncol Lett. 2019;17:3399–3405.
    57. Tylman M, Sarbinowski R, Bengtson JP, Kvarnström A, Bengtsson A. Inflammatory response in patients undergoing colorectal cancer surgery: the effect of two different anesthetic techniques. Minerva Anestesiol. 2011;77:275–282.
    58. Grande M, Tucci GF, Adorisio O, et al. Systemic acute-phase response after laparoscopic and open cholecystectomy. Surg Endosc. 2002;16:313–316.
    59. Colley CM, Fleck A, Goode AW, Muller BR, Myers MA. Early time course of the acute phase protein response in man. J Clin Pathol. 1983;36:203–207.
    60. Wang K, Wu M, Xu J, et al. Effects of dexmedetomidine on perioperative stress, inflammation, and immune function: systematic review and meta-analysis. Br J Anaesth. 2019;123:777–794.
    61. McGain F, Muret J, Lawson C, Sherman JD. Environmental sustainability in anaesthesia and critical care. Br J Anaesth. 2020;125:680–692.
    62. Zhang Y, Shan GJ, Zhang YX, et al.; First Study of Perioperative Organ Protection (SPOP1) investigators. Propofol compared with sevoflurane general anaesthesia is associated with decreased delayed neurocognitive recovery in older adults. Br J Anaesth. 2018;121:595–604.
    63. Mei X, Zheng HL, Li C, et al. The effects of propofol and sevoflurane on postoperative delirium in older patients: a randomized clinical trial study. J Alzheimers Dis. 2020;76:1627–1636.
    64. Alhayyan A, McSorley S, Roxburgh C, Kearns R, Horgan P, McMillan D. The effect of anesthesia on the postoperative systemic inflammatory response in patients undergoing surgery: a systematic review and meta-analysis. Surg Open Sci. 2020;2:1–21.
    65. Wan Y, Xu J, Ma D, Zeng Y, Cibelli M, Maze M. Postoperative impairment of cognitive function in rats: a possible role for cytokine-mediated inflammation in the hippocampus. Anesthesiology. 2007;106:436–443.
    66. Hu J, Feng X, Valdearcos M, et al. Interleukin-6 is both necessary and sufficient to produce perioperative neurocognitive disorder in mice. Br J Anaesth. 2018;120:537–545.
    67. Silbert BS, Evered LA, Scott DA. Incidence of postoperative cognitive dysfunction after general or spinal anaesthesia for extracorporeal shock wave lithotripsy. Br J Anaesth. 2014;113:784–791.
    68. Deiner S, Baxter MG, Mincer JS, et al. Human plasma biomarker responses to inhalational general anaesthesia without surgery. Br J Anaesth. 2020;125:282–290.
    69. Borchers F, Spies CD, Feinkohl I, et al. Methodology of measuring postoperative cognitive dysfunction: a systematic review. Br J Anaesth. 2021;126:1119–1127.
    70. Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1. Introduction - GRADE evidence profiles and summary of findings tables. J Clin Epidemiol. 2011;64:383–394.
    Copyright © 2021 International Anesthesia Research Society