PDF
Citation
Print
INTRODUCTION
Despite advances in surgical techniques, intraoperative bleeding during liver surgery remains a concern. Intraoperative blood loss is considered a cause of perioperative complications in patients undergoing liver surgery [1]; hence, several operative and non-operative strategies are applied to reduce the risk of bleeding. The strategies include intermittent clamping the portal vein and hepatic artery (Pringle manoeuvre) [2], sophisticated liver dissection techniques, use of stapling devices, ultrasonic dissectors, bipolar cautery, and hydrodissectors for parenchymal transection [3], and maintaining a low central venous pressure (CVP) during parenchymal transection [4,5].
Although the learning curve is steep and there are technical challenges, laparoscopic liver resection has become more popular in the past decade even for lesions in anatomically challenging segments [6,7]. However, the application of haemostasis techniques may be more challenging during laparoscopic liver resections compared with the open approach [8]. It has been shown that low CVP can reduce intraoperative bleeding during liver resection [5]; however, the impact of low CVP on intraoperative bleeding, specifically in laparoscopic liver resection is not known. Therefore, this systematic review and meta-analysis aimed to compare the outcomes of low CVP to standard CVP during laparoscopic liver resection.
MATERIALS AND METHODS
The design of the study was according to the Preferred Reporting Items for systematic reviews and Meta-Analyses statement standards [9], and followed a predefined protocol.
Criteria for eligibility
Study design: Comparative cohort studies and randomized controlled trials (RCTs) that compared low CVP with standard CVP in patients undergoing laparoscopic liver resection were of interest to this review.
Population: Adult patients undergoing laparoscopic liver resection for any indications were eligible. The liver resection procedures of interest comprised both anatomical and non-anatomical resections.
Intervention and comparison: Low CVP, defined as pressure ≤ 5 cmH2O, was the eligible intervention and standard CVP, defined as pressure > 5 cmH2O, was the eligible comparison.
Outcomes: Intraoperative blood loss was the primary outcome. The secondary outcomes comprised the need for blood transfusion, mean arterial pressure, operative time, Pringle time, and total complications.
Literature search strategy
Using proper keywords, thesaurus headings, and search limits, a literature search strategy was constructed by two separate authors who had adequate experience in evidence synthesis (Appendix 1). There were no language restrictions. The date for the last search was 15 July 2023. The electronic sources searched comprised MEDLINE, CINAHL, Scopus, International Standard Randomised Controlled Trial Number Registry, World Health Organization International Clinical Trials Registry, CENTRAL, ClinicalTrials.gov, the European Association for Grey Literature Exploitation, and System for Information on Grey Literature. The reference lists of the relevant original studies and reviews were also evaluated.
Selection of the eligible studies and data collection
Two separate authors reviewed the titles and abstracts of the articles, obtained the full texts of potentially eligible articles, and included the eligible studies. The following relevant data from the eligible studies were recorded on an electronic data collection sheet:
• First author’s name
• Publication year
• Journal name
• Study design
• Included population
• Study sample size
• Definitions of low and standard CVP
• Outcomes
Any disagreements about study selection or data collection were resolved by a third author when required.
Risk of bias assessment
The risk of bias (RoB) of RCTs was judged using the Cochrane RoB tool [10], and the RoB of observational studies was evaluated using the Risk Of Bias In Non-Randomized Studies of Interventions (ROBINS-I) tool [11]. Two independent authors conducted the RoB assessment and a third author resolved the disagreements when required.
Data analysis
Review Manager 5.4 software and trial sequential analysis (TSA) 0.9.5.5 Beta software were used. Random effects modelling was used to compute mean difference (MD) with 95% confidence level for continuous variables and odds ratio (OR) with 95% confidence for dichotomous variables. An individual patient was the unit of analysis and we based the analyses on intention to treat data. The I2 using the Cochran Q test (χ2) was calculated to measure statistical heterogeneity and was interpreted as below:
• I2 0%–25% suggests low heterogeneity
• I2 25%–75% suggests moderate heterogeneity
• I2 75%–100% suggests high heterogeneity
When an outcome was reported by ten studies, we aimed to construct funnel plots for publication bias risk assessment. TSA was modelled to explore the possibility of type I and type II errors in the primary outcome analysis as long as it was adequately reported (by at least five RCTs). The possibility of type I and II errors was determined using the O’Brien-Fleming α-spending function and futility boundaries, respectively.
Additional analyses
Leave-one-out analysis and independent meta-analysis of studies with low RoB were planned as sensitivity analyses of the primary outcome.
Certainty of evidence
The GRADE system was followed to determine the certainty of evidence for each outcome [12].
RESULTS
The literature search yielded 277 articles; 266 studies were excluded due to irrelevance. After full-text review, three more articles were excluded as they were systematic reviews (Fig. 1). The remaining eight articles [13-20] met the eligibility criteria of this study. The eligible studies included 7 RCTs and 1 retrospective cohort study, enrolling a total of 682 patients; 342 in the low CVP group and 342 in the standard CVP group (Table 1).
Risk of bias assessment
The outcomes of RoB assessment based on the aforementioned tools are shown in Fig. 2.
Primary outcome
Intraoperative blood loss: Examination results of 682 individuals (8 studies) demonstrated that low CVP reduced intraoperative blood loss (MD, –193.49 mL; 95% confidence interval [CI], –339.86 to –47.12; p = 0.01). The heterogeneity (statistical) was high (I2 = 99%, p < 0.001) and the GRADE system suggested moderate certainty (Fig. 3). The information size calculated via TSA (950 patients) was not achieved; however, Z-curve intersected the conventional and alpha-spending boundaries in favour of low CVP, hence the possibility of type I and II errors was minimal and the conclusion was robust (Fig. 4).
Secondary outcomes
Need for blood transfusion: Examination results of 416 individuals (4 studies) demonstrated comparable need for blood transfusion between low CVP and standard CVP (OR, 0.54; 95% CI, 0.28–1.03; p = 0.06) (Fig. 3). The heterogeneity (statistical) was low (I2 = 0%, p = 0.85) and the GRADE system suggested moderate certainty.
Mean arterial pressure: Examination results of 364 individuals (3 studies) demonstrated comparable mean arterial pressure between low CVP and standard CVP (MD, –1.55 mm Hg; 95% CI, –3.85–0.75; p = 0.19) (Fig. 3). The heterogeneity (statistical) was low (I2 = 0%, p = 0.75) and the GRADE system suggested moderate certainty.
Pringle time: Examination results of 404 individuals (3 studies) demonstrated comparable Pringle time between low CVP and standard CVP (MD, –0.99 minutes; 95% CI, –5.82–3.84; p = 0.69) (Fig. 3). The heterogeneity (statistical) was high (I2 = 96%, p < 0.001) and the GRADE system suggested moderate certainty.
Operative time: Examination results of 454 individuals (4 studies) demonstrated comparable operative time between low CVP and standard CVP (MD, –16.38 minutes; 95% CI, –36.68–3.39; p = 0.11) (Fig. 3). The heterogeneity (statistical) was high (I2 = 95%, p < 0.001) and the GRADE system suggested low certainty.
Total complications: Examination results of 274 individuals (4 studies) demonstrated comparable risks of total complications between low CVP and standard CVP (OR, 1.92; 95% CI, 0.97–3.80; p = 0.06) (Fig. 3). The heterogeneity (statistical) was low (I2 =0%, p = 0.66) and the GRADE system suggested moderate certainty.
DISCUSSION
We conducted a systematic review to compare the outcomes of low CVP and standard CVP during laparoscopic liver resection. Analysis of 8 studies including 682 patients showed that low CVP reduced intraoperative blood loss during laparoscopic liver resection; however, it did not have any effects on the need for blood transfusion, mean arterial pressure, Pringle time, operative time, or total complications. Sensitivity analysis suggested consistency of the results for intraoperative blood loss and TSA suggested that meta-analysis for intraoperative blood loss was not subject to type I or II errors. The GRADE system suggested moderate certainty.
For the first time in the literature, the current study assessed the impact of low CVP on intraoperative bleeding, specifically during laparoscopic liver resection. However, the results of the current study can be compared with the results of previous reviews, which did not limit the included population to laparoscopic liver resection. Liu et al. [5] conducted a systematic review including patients who had open or laparoscopic liver resection and concluded that low CVP is effective in reducing blood loss during liver resection. In another study, Hughes et al. [21] concluded that low CVP reduced intraoperative bleeding during open liver resection. Our findings are consistent with those of the aforementioned reviews.
The reduced intraoperative blood loss associated with low CVP during laparoscopic liver resection can be simply explained. It is well-recognized that venous bleeding is the main source of bleeding during liver resection, especially when hepatic inflow is reduced via Pringle manoeuvre or hepatic hilum occlusion [22,23]. Considering that CVP is directly related to the hepatic sinusoidal pressure, reducing the pressure in the inferior vena cava will reduce the former resulting in a drop of the pressure in hepatic sinusoids hence reducing intraoperative blood loss [22,23]. The findings of the current study support the above mechanism.
The current study is subject to the following limitations. Although most of the included studies had a randomized design, the included population characteristics, detail of liver resections, and detail of randomizations were poorly reported. Moreover, the available evidence is limited to studies from the same country, which may affect the generalisability of the findings. Consequently, the available data was not adequate for performing meta-regression analysis or subgroup analyses based on the baseline characteristics of the included population and type of liver resection; hence, confounding bias and selection bias cannot be excluded. Laparoscopic insufflation pressure may help to reduce intraoperative blood loss when it exceeds CVP pressure during laparoscopic liver resection. Therefore, variation in insufflation pressure among the included studies and between the treatment arms may be an important confounding factor. The insufflation pressure used during laparoscopic pressure was poorly reported by the included studies; consequently, it was not possible to perform meta-regression analysis or subgroup analyses based on insufflation pressure; this can be considered as another potential source of confounding bias. Although the intraoperative blood loss as the primary outcome was well-reported by the included studies, the secondary outcomes were not reported by all of the included studies; hence, our secondary outcomes results are less robust. The statistical heterogeneity was high for the primary outcome; nevertheless, the findings remained consistent through sensitivity analyses and TSA. Moreover, we downgraded evidence certainty due to high statistical heterogeneity. Because we included less than 10 studies, we were not able to assess publication bias risk.
CONCLUSIONS
While the available evidence may be subject to confounding bias and selection bias, the best available evidence suggests that low CVP may reduce intraoperative bleeding during laparoscopic liver resection (moderate certainty); however, this may not translate into shorter operative time, shorter Pringle time, or less need for blood transfusion. RCTs with larger sample sizes will provide more robust evidence. The results of the current study provide a robust basis for power analysis and hypothesis synthesis for future randomised trials.
XML Download