Evaluating and developing sealants for the prevention of pulmonary air leakage: A systematic review of animal models

Abstract

German

Spanish

French

Sealants may provide a solution for pulmonary air leakage (PAL), but their clinical application is debatable. For sealant comparison, standardized animal models are lacking. This systematic review aims to assess methodology and quality of animal models for PAL and sealant evaluation. All animal models investigating lung sealing devices (e.g., staplers, glues, energy devices) to prevent or treat PAL were retrieved systematically from Embase, Pubmed and Web of science. Methodological study characteristics, risk of bias, reporting quality and publication bias were assessed. A total of 71 studies were included (N = 75 experiments, N = 1659 animals). Six different species and 18 strains were described; 92% of experiments used healthy animals, disease models were used in only six studies. Lesions to produce PAL were heterogenous, and only 11 studies used a previously reported technique, encompassing N = 5 unique lesions. Clinically relevant outcomes were used in the minority of studies (imaging 16%, air leak 10.7%, air leak duration 4%). Reporting quality was poor, but revealed an upward trend per decade. Overall, high risk of bias was present, and only 18.7% used a negative control group. All but one study without control groups claimed positive outcomes (95.8%), in contrast to 84.3% using positive or negative control groups, which also concluded equivocal, adverse or inconclusive outcomes. In conclusion, animal studies evaluating sealants for prevention of PAL are heterogenous and of poor reporting quality. Using negative control groups, disease models and quantifiable outcomes seem important to increase validity and relevance. Further research is needed to reach consensus for model development and standardization.

Keywords

Animal model ethics and welfare experimental design organisms and models refinement surgery techniques

Introduction

Pulmonary air leakage (PAL) remains an important cause of complications following pulmonary resection in modern thoracic surgery. Air leaks (AL) persisting for more than five days are generally considered to be prolonged (pPAL), and estimates are that 5.6–30% of patients are affected.¹ As a consequence of pPAL, post-operative complications (30% versus 18%) and mortality (odds ratio (OR) 1.90, 95% confidence interval (CI) 1.42, 2.55) are increased.^2–4 Hospital stay is extended four to eight days and patients are subject to more re-interventions and re-admissions.^5–9 As a result, complications associated with pPAL increase both hospital and societal costs.^8,9 To mitigate the impact of pPAL, lung sealing devices (LDs) have been proposed for intraoperative use.

LDs are a heterogeneous group of devices and implants (e.g., staplers, glues, energy devices) to establish an aerostatic seal of the resected lung. A large number of LDs have been evaluated with positive results in animal studies. However, currently, only 42% of surgeons find the evidence for routine use of LDs (with the exception of staplers) compelling.¹⁰ To summarize the evidence based on several systematic reviews and meta-analysis, there appears to be a positive effect on PAL, but impact on clinically relevant outcomes such as complications and length-of-stay are debatable.^11–14 Furthermore, there is considerable doubt regarding cost-effectiveness.¹⁵ Another important concern is the lack of clinical trials performing head-to-head comparisons of different LDs.¹³ So, although there is some evidence for a number of LDs, none seems to work perfectly and the search for better solutions for pPAL continues, requiring a standardized and valid pre-clinical work-up.¹⁰

The discrepancy between positive results derived from animal studies and the debatable clinical benefits from LDs is noteworthy. In this regard, valid animal models are indispensable for direct comparison or novel product development purposes. However, the current literature of in vivo studies investigating LDs is heterogenous and there are no widely accepted standards for research of pPAL in animal models.^16–21 This is in contrast to animal studies in different medical fields such as gastro-intestinal anastomosis.²² This makes standardized pre-clinical research into novel and existing LDs especially difficult. Considering that conduction of animal research with poor methodological quality (e.g., lack of randomization and blinding) or invalid model designs (e.g., healthy animals and lack of clinically relevant outcomes) may be one of the reasons for insufficient performance of LDs in clinical practice, further investigations are necessary.²³

Live animals are still required in this field of research due to the complex mechanisms involved, including foreign body reactions, coagulation cascades, local tissue homeostasis and pleural mechanisms in spontaneously breathing mammals, that cannot yet be replaced by animal-free models.^24–26 Due to the ethical and societal concerns associated with animal experiments, it is especially important to only perform research of the highest internal quality, in the best model with the highest external validity and in the most refined manner possible.²³ In order to identify and improve on the described methodological problems, animal systematic reviews are recommended in the translational research phase.²⁷

The aim of this systematic review is to provide a comprehensive overview of the methodological characteristics of previous animal models used for the evaluation of LDs for the treatment of PAL, and provide recommendations for future research in this field. Furthermore, the risk of bias and quality of reporting will be assessed, providing a baseline for quality improvements in this field.

Methods

Search strategy

A systematic review study protocol was registered in the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY) under registration number 202270003. A search strategy with three components was devised to identify all animal studies describing the evaluation of LDs as a treatment of parenchymal PAL. In short, one component including terms related to pulmonary surgery and post-operative air leakage and a second component with terms related to LDs were combined using the animal studies search filter designed by Hooijmans.²⁸ Key Mesh terms included: ‘Pulmonary surgical procedures,’ ‘Pneumothorax,’ ‘Respiratory tract fistula,’ ‘Pneumonectomy,’ ‘Tissue adhesives’ and ‘Fibrin tissue adhesive.’ This search strategy was deployed in Pubmed, EMBASE and Web of Science on 9-2021 (see Supplemental material A for complete database entries). No language or publication date restrictions were applied to the literature search.

Identification of relevant articles

Citation information from each of the database searches was collected in Endnote (version 20), and duplicates were automatically removed. Title and abstract screening was then performed on the collected references in Endnote citation manager by one investigator (BH or JE); in case of doubt, the full-text was retrieved for screening. Full-text retrieval was then done through (a) university access, (b) online searches and (c) with the help of our university library contacting national and international universities. Full-text screening of possible relevant studies was performed by consensus between two investigators (BH and SP/JE). In case of non-English papers, the papers were translated using Google Translate for screening purposes. If this translation was insufficient, papers were screened with the help of native speakers. Reference lists of included English studies were screened by one author (BH or JE) to identify potential additional relevant articles. Inclusion criteria were (a) in vivo study design in mammals, (b) model of parenchymal PAL (small bronchioles within lesions may be included), (c) sealing of the AL with a surgical LD and (d) assessing aerostatic efficacy of the LD. The following exclusion criteria were considered: (a) LD is only used to seal a large bronchus or trachea (such as large segmental or lobar bronchi), (b) LD is a non-surgical intervention (such as bronchoscopy, pleurodesis, thoracic drainage) and (c) studies only evaluating the hemostatic or biocompatibility characteristics of the LD under investigation. Any article form other than a complete research paper (e.g., conference abstracts) were also excluded. All included non-English full texts were translated using Google Translate, and two Russian papers were translated by a native speaker. In cases where the quality of the translation was not sufficient to reliably perform data-extraction or risk of bias grading, data extraction and quality grading was performed by an academic native speaker (N = 4 Japanese studies) and the help of one researcher (BH).

Data extraction and quality assessment

In order to generate an overview of the methodologies used in all animal studies, a wide range of characteristics relating to the animal model (age, sex, strain, weight, disease model), air leak model (surgical characteristics, defect descriptions, baseline leak measurements), lung sealing and outcomes (acute sealing testing, histology, bursting pressure, air leak, imaging, macroscopy, adverse events) were extracted from each study. A complete overview is provided in the table in Supplemental material B.

For assessment of trends in reporting quality, a custom score was calculated (18 points max), that was derived from a similar score used by Yauw et al. in their animal systematic review, designed for surgical animal studies.²⁹ We chose items that were most appropriate with regard to experimental investigations of lung sealants, and objectively reproducible when scoring (i.e., items do not require any form of reviewer judgement such as in risk of bias scoring). This score was partly based on items from the ARRIVE Guidelines (strain, sex, age, weight, housing, anesthesia, analgesia post-operatively, antibiotic prophylaxis, sterility during surgery, ethical statement, data access statement, registration of protocol) and additional items added relevant to the study of lung sealants (type of incision to gain access to the lungs, location of defect right or left lung, location of defect lobe used, creation of the lesion on a static lung, application of the sealant on a static lung, defect dimensions).^29,30 The lung was considered static in case the area of interest was not being ventilated during relevant parts of the procedure (e.g., using a clamp, single-lung ventilation, continuous positive airway pressure or ventilatory arrest). Each item was scored as ‘1’ if reported, and ‘0’ if not reported or not applicable. All variables were extracted by one author (BH), and completely checked for mistakes by a second author (SP). Data was extracted only from text and tables.

For study quality assessment, the SYRCLE Risk of bias tool (RoB) was used, which consists of bias assessment in six domains, specifically designed for use in animal systematic reviews. Items 1–8 on this tool were assessed using signaling questions, which were answered as ‘yes,’ ‘no’ or ‘unclear’ (see Supplemental material for signaling questions used). Thereafter, items were scored as ‘high,’ ‘unclear’ or ‘low’ risk of bias.³¹ In the specific case of poorly reported studies, often no mention is made regarding a signaling question, and we considered the subsequent ‘probable no’ answer similarly as ‘no’, generally leading to a high risk of bias grading. Furthermore, an additional six questions related to internal validity were formulated (randomization, blinding, power-calculation, use of positive and negative control groups and industry funding) and answered with ‘yes,’ ‘no’ or ‘unclear.’ Publication bias is qualitatively assessed based on authors conclusions regarding the efficacy of the LDs (positive, equivocal, adverse, inconclusive). Funnel plot analysis was not feasible in this regard due to high heterogeneity of interventions and outcomes and poor outcome reporting quality. All quality assessment was initially done by reaching consensus between two authors (BH and SP), except for the papers read by the Japanese native speaker (N = 4).

Data analysis

All study characteristics were gathered and analyzed using IBM SPSS Statistics for Windows, Version 27.0 (IBM Corp., Armonk, New York). GraphPad Prism (version 9) was used for graphical presentation of data. Methods of animal models were assessed in a semi-quantitative analysis, describing model characteristics with descriptive statistics. Trends due to date of publishing were investigated using Pearson’s correlation. Due to the large heterogeneity between included studies, no meta-analysis or meta-regression was performed.

Results

Description of included studies

A total of 71 papers (Supplemental material C) were included in this review, describing 75 experiments (Figure 1). Three papers were identified by reference list screening but, upon inspection, were also present in the original search and missed during screening. Only one paper was identified completely separately from the initial literature search and reference list screening. Two articles first appeared to meet the inclusion criteria, but were excluded due to very poor descriptions of methodology and results, making reliable data-extraction and risk of bias grading impossible. Study characteristics were extracted per experiment and risk of bias grading was performed per paper. Most experiments were performed after 1990, with 17 experiments dating from the 35-year period between 1955 and 1989 (Figure 2). A total of 1659 animals were used (mean, 23; SD, 29 per experiment; N = 3 missing data) with a downward trend per decade (R = –0.140, P = 0.241).

Figure 1.

Identification eligible of studies in databases and registries.

Figure 2.

General study characteristics and animal models used. (a) Study continent of origin grouped per decade. Dotted bar, estimate for this decade based on linear extrapolation and (b) Animal model used grouped per decade.

Quality assessment

Study quality assessment is shown in Figure 3. The custom reporting quality score revealed an upward trend through the decades (R = 0.539, P < 0.001). Items most frequently left unreported were registration of protocol (100%), data access (96%), housing details (89%), post-operative analgesia (89%), perioperative antibiotic prophylaxis (80%), sterile surgery (76%), description of static lung state during defect creation (75%), ethical approval statement (63%), description of static lung state during LD application (61%) and anatomical location of lesion (53%).

Figure 3.

Study quality assessment. (a) Risk of bias assessment, based on the SYRCLE risk of bias tool. (b) Custom reporting quality score, revealing an upward trend per decade (R = 0.539; P < 0.001). (c) Additional risk of bias questions and (d) Publication bias assessed by collecting authors conclusions, grouped by control groups used.

Based on the SYRCLE RoB bias tool, no studies were considered at low risk of bias and, in all domains, the majority of studies were graded as being at high risk of bias (Figure 3; Supplemental material). In the additional questions, it was noted that negative control groups were used in only 18.7% of studies. Regarding publication bias, all but one of the studies without any control groups claimed positive outcomes, in contrast to the 15.7% using either positive or negative control groups, that also concluded equivocal, adverse or inconclusive outcomes.

Animal models

Included experiments had the aim of showing either aerostatic efficacy (30.7%) or both aerostasis and biocompatibility (69.3%). In 24% of studies, the animals were sacrificed immediately after surgery. Use of six different species was described (canine 36%, porcine 25.3%, rats 18.7%, rabbits 13.3%, ovine 5.3% and mice 1.3%). Strain was further defined in 68% of studies, revealing a heterogenous pool of animal models (N = 18). Of these, most used were Beagle dogs (23.5%), New Zealand white rabbits (15.7%), Wistar rats (13.7%) and Mongrel dogs (11.8%). Sex, weight and age were described in 48.7%, 39.7% and 74.4% of studies, respectively. Both sexes were only used in 11.1% of studies that specified sex, and animals used were frequently relatively young (64.7% <1 year). Mainly healthy animals were used (92%), and disease models used comprised athymic rats (N = 3), coagulopathy (N = 2) and emphysema (N = 1).^32–36

Air leak models

The surgical approach for defect creation consisted of a thoracotomy (82.7%) in most cases, and the left lung (40%) was used more frequently than the right lung (32%). The precise anatomical location of the lung defect was reported in 46.5% of experiments. If described, a heterogenous division of locations was seen (Figure 4), with a median number of defects of one per animal (interquartile range (IQR), 1; range, 14). The defect models used were even more heterogenous, with only N = 5 defects being used in more than one experiment (combined N = 11 experiments). Exact defect dimensions were described in 68% of experiments, and these lesions were made at a varying depth (Figure 4). However, only 25.3% of authors describe the creation of lesions on a static lung, for example, with a clamp applied to the lung parenchyma or at a static ventilatory pressure, making the actual number of lesions that can accurately be reproduced lower.

Figure 4.

Pulmonary defect models used. (a) Surgical technique and lung used. (b) Defect type used, grouped by the description of static lung state during defect induction. (c) Precise anatomical defect location used and (d) Defect depth, grouped by presence or absence of a complete description of defect dimensions. *Defect depth not applicable (N/A) (e.g., wedge resection or only removal of pleura).

At baseline, hemostatic measures other than the applied lung sealant were described in 20% of studies. With respect to potential confounding factors, assessment of baseline air leak was described infrequently: 40% of studies included some descriptive indication of baseline air leak (e.g., ‘Leaking of blood and air were confirmed’ or ‘Air leak confirmed by bubble formation’) and formal methods with quantifiable outcome measures were used in 15 (20%) studies.^35,37 Quantifiable measures included a water submersion test with visual grading, bursting pressure/minimal leaking pressure or actual air leak amount (Figure 5).^17,38–41

Figure 5.

Baseline defect and sealing characteristics. (a) Hemostasis techniques used. (b) Lung sealant type applied, grouped based on the description of static lung state during sealant application. (c) Baseline air leak measurements performed and (d) Acute lung sealing tests performed.

Lung sealing and outcomes

A wide range of LDs, grouped according to application category in Figure 5, have been evaluated in animal experiments. Regarding methodology, 38.7% describe application on a static lung. In 50.7% of experiments, an acute sealing test directly after application was performed, either by water submersion testing or application of a threshold ventilatory pressure (maximum 45 cm H₂O). Remarkably, one paper describes pressure resistance of up to 350–400 cm H₂O, which seems unlikely for the canine lungs to have withstood.⁴² LDs were evaluated using only positive pressure ventilation in 22.7% of experiments, and using both mechanical ventilation and physiological breathing in 74.7% of experiments. The most evaluated late outcomes included were histology (77.3%), adverse events (62.7%, such as a post-operative pneumothorax, infections or death), macroscopic observations (52%) and bursting pressures (42.7%). Late evaluation was usually performed at a median follow-up of 30 days (IQR, 46 days). A clinically relevant outcome including pneumothorax detection with imaging (16%), air leak (10.7%) and air leak duration (4%) was measured in a small proportion of experiments.

Discussion

For the purpose of evaluating LDs as a treatment for PAL, no standardized animal model has been described to date. This is supported by this systematic review, which revealed high heterogeneity regarding methodological characteristics and outcome measures between all previous animal studies. High risk of bias and poor reporting quality complicate further standardization of models. Translatability of results may be further impaired by lack of quantifiable measures of PAL and lack of disease models in the previous animal studies. As such, including a negative control group will help to increase study validity by assessing the natural healing course of a lesion. Confirming that a lesion results in PAL and not immediate closure from intrinsic healing will help reduce publication bias.

Interpretation and comparison with literature

To the best of our knowledge, this is the first systematic review of literature performed on the use of LDs for treatment of PAL in preclinical animal studies. Nevertheless, the methodological shortcomings illustrated by this systematic review are a common finding in other preclinical systematic reviews, including lack of randomization, allocation concealment, blinding, low reporting quality and the presence of a publication bias.^27,29 Another common problem is heterogeneity due to lack of standardized protocols, as was also illustrated in the preclinical systematic review on intestinal anastomosis.²⁹ Reducing this heterogeneity can allow for better meta-analysis and investigation of external validity by comparison of pooled preclinical findings with clinical data.²⁹ A recent review article by van der Worp and coworkers described common causes of reduced external validity to be due to insufficient use of disease models, use of a homogenous group of animals, using only male or female animals, insufficient similarity between clinical and experimental disease and difference in outcome measures.²³ All these factors were also identified to be lacking in the included animal studies. From this perspective, systematic appraisal of preclinical literature can be beneficial, allowing rationalized recommendations for future research, as will be further discussed.²⁷

The current preclinical literature does not provide sufficient evidence to conclude standards for making a lesion that results in relevant and persistent PAL. While many different lesions have been investigated, appraisal of their validity is impossible due to lack of negative control groups or quantifiable outcome measures. Moreover, there is evidence that healthy animal lungs may possess strong intrinsic healing mechanisms that can provide fast sealing of relative large lung lesions. For example, removal of the entire parietal pleural surface at 3 mm depth in the healthy dogs was not associated with air leak complications.⁴³ Another study performed segmental resections with a negative control group, leaving the raw parenchymal area of 9–15 cm² untreated.⁴⁴ No air leak complications were seen due to this negative control lesion, making interpretation of aerostatic efficacy of evaluated treatments impossible.⁴⁴ Yet another study with especially large square defects (9 × 9 × 0.5 cm) in healthy dogs illustrated intrinsic healing in four out of eight negative control animals within 24 h, the same proportion as in the fibrin glue treated group.¹⁷ In our own work with healthy sheep, we have observed similar rapid intrinsic healing of parenchymal lesions (unpublished data).

Considering these intrinsic healing mechanisms, a valid model that results in relevant and persistent PAL is required. In the absence of standardized validated models, a negative control group should be included to confirm the presence of PAL when left untreated. For standardized model development, the current literature offers several leads. Several studies have shown that non-anatomical partial resections of the lung are associated with air leaks and air leak complications in healthy animals.^38,45,46 The leakage capabilities of such lesions are potentially related to the laceration of bronchioli. In the study by Ranger and colleagues, bronchioli of 1.5–3.0 mm in diameter were intentionally included, observing significant air leaks over the 24-h observation period in the negative control group.³⁸

Secondly, from a clinical perspective, a disease model to increase the odds for pPAL could be considered. The healthy animals used in most pre-clinical studies differ from the diseased patients who are at risk of pPAL. In clinical practice, patients undergoing pulmonary resections, especially those at risk for pPAL, are often older, have a history of smoking and chronic obstructive pulmonary disease, take medications (anticoagulation, antiplatelets, steroids) and might have had chemo- and radiotherapy prior to surgery.^1,47–49 In contrast, laboratory animals are often younger, healthy and have no prior history of disease. Disease models impairing immune function, lung quality or coagulation mechanisms were used in only a minority of studies.^32,35,36 This difference might impact the healing course of lung lacerations and modify LD results accordingly.

A promising disease model might be the emphysema model. Gika et al. reported inferior bursting pressure results for all LDs under investigation applied in their emphysema model compared with the healthy animal model.³⁶ A drawback of this model is that the induction of emphysema by bronchoscopic elastase instillation is difficult to standardize.³⁶ Balakrishnan described a disease model that is easy to standardize, based on heparinization.³⁵ Such a model might reduce intrinsic healing mechanisms, but no negative control group or quantifiable outcomes were used to confirm that this method enhances risk of PAL.³⁵ In conclusion, the use of disease models seems promising; however, it could come at a cost of less standardization, greater study complexity and reduced animal welfare. Use of disease models should be weighed ethically in the light of the aforementioned uncertainties and experimental aims.

Considering further standardization of animal models, reproducible lesions should be ensured across animal experiments, and the precise method of induction and lung expansion state should be specified. The lung changes throughout the respiratory cycle with a volumetric strain of 20–50%.⁵⁰ Therefore, a lung defect should be created on a set inflation state for valid and reproducible results (e.g., empty lung or 10 cm H₂O positive pressure). Furthermore, when creating multiple lesions, the influence of anatomical location on PAL should also be considered. For instance, a clinical association has been described between upper lobectomies and occurrence of PAL in humans.^49,51 Mechanistically, this could possibly be explained by the effects of gravity on the intrapleural pressure gradient, which has a different orientation in quadrupedal animals.^52,53 The influence of species effects on PAL from a lung lesion and intrinsic healing in animal models is currently unknown.

Recommendations for future research

Because of the lack of standardized preclinical models known to reliably result in PAL, the inclusion of negative control groups is needed to confirm model validity. Ultimately, PAL findings need to be studied across varying lesions, species, anatomical locations and disease models and compared systematically with clinical data to determine the preclinical model with the highest external validity. Previous literature indicates that parenchymal lesions in healthy animal might seal intrinsically, whereas lesions with bronchioli are capable of PAL, but this should be investigated further.^17,38,43 When creating lesions, the precise conditions (e.g., lung inflation state, dimensions, location) should be documented to allow for standardization. Finally, specific disease models such as pulmonary emphysema or coagulopathy might be used to enhance translational value.^35,36

Air leakage should be quantified objectively, both as a baseline confounder and outcome parameter and an air leak should be of adequate size, at least comparable with those of clinical cases where a lung sealant might be indicated for use. Recent consensus guidelines advice to classify air leaks as mild (<100 ml/min), moderate (100–400 ml/min) and severe (>400 ml/min).⁵⁴ In other studies, intraoperative PAL >500 ml/min or >9.5% of the inspiratory tidal volume have been found to be predictive of pPAL.^55,56 Similar classification might also apply to larger animals such as sheep and pigs. Classifications of air leakage that are specified or validated for animal species are not available and should be developed further.

Risk of bias can be mitigated by use of randomization, allocation concealment and use of negative control groups. Studies should also be sufficiently powered to detect meaningful outcomes. Although increasing the power of an animal model initially might seem to increase the number of animals needed, it will eventually reduce the amount of animals needed by lowering the need to perform duplicate models.⁵⁷ To further improve the methodologies of animal experiments, Delphi consensus guidelines made by an international board of experts could be used to standardize models and improve translational value.^22,29 To improve internal validity, researchers are encouraged to use the ARRIVE guidelines when designing and publishing animal experiments (see Table 1 for summary of recommendations).³⁰

Table 1.

Summary of recommendations for future research.

Topic	Author recommendations	Knowledge gap
Animal model	• Consider large animals for aerostatic efficacy testing (similar biomechanics)	• Species-specific intrinsic pulmonary healing differences • Bipedal versus quadrupedal and anatomical location effect-modification
Disease model	• Develop disease model that enhances pulmonary air leak and reduces intrinsic healing	• Validity of heparinized disease model³⁵ • Validity of emphysema disease model³⁶
Air leak model	• Create lesions on static lung with defined pressure conditions • Consider large lesions in healthy animals (extended parenchymal or inclusion of bronchioles 1.5–3.0 mm)^17,38 • Aim for air leak >400–500 mL/min (large animals) or >9.5% of tidal volume^54–56	• Dose-response relationship between different lesion types/sizes on air leak (study inter- and intraspecies) • Translation factor of animal versus clinically relevant air leaks
Outcome measures	• Ensure baseline measurements (lesion size, air leak, bronchioles, coagulation times) to control between groups • Consider similar outcome measures to clinical practice (air leak size and duration, time until drain removal)	• Translational value of animal outcome measure to clinical practice
Reduction and refinement	• Consider multiple lesions per lung (large animals) to increase power (e.g., animals as own control) and calculate required sample size beforehand • Use multimodal analgesia principles from clinical thoracic surgical practice • Use transcostal sutures to reduce post-operative pain⁵⁸	• Mutual influence with multiple lesions per lung (e.g., through pneumothorax, empyema)
Risk of bias	• Use negative controls to adjust for intrinsic healing effects • Use randomization and blinding principles • Adhere to ARRIVE guidelines³⁰	• Gather panel of experts to specify and improve recommendations

Limitations

Because of the high number of studies lacking a control group and quantifiable outcome measures, publication bias is difficult to assess, and standard techniques such a funnel plots were insufficient. Therefore, an analysis of author conclusions was performed as a surrogate outcome for publication bias.²⁹ This analysis showed that publication bias is likely, as the majority of studies reported positive conclusions.

To ensure a valid assessment of risk of bias, this was performed by two investigators independently, after which consensus was reached. Our assessment may be a conservative estimation of the overall study quality and not fully represent the true risk of bias, because there was also a low reporting quality in most studies. Cases were a specific aspect of study design was not reported, were generally treated as high risk of bias (e.g., no mention of randomization), but some uncertainty remains whether the experiments were actually highly biased or poorly reported.⁵⁹ Either way, problems with poor reporting or risk of bias are evident in the current body of literature, posing ethical concerns when using laboratory animals and contributing to research waste and poor reproducibility. However, due to the underestimation of the quality of evidence, the results should thus be interpreted with appropriate caution.

The custom reporting quality scoring has two associated limitations. First of all, not applicable items were not accounted for in overall scoring (e.g., sterility or antibiotic prophylaxis in short terminal experiments), which might lead to an underestimation of the reporting quality. Secondly, no complete scoring of all items on the ARRIVE guidelines was performed, which may have further increased the relevance of the reporting quality scoring in the context of animal studies.

Conclusions

This systematic review underlines the methodological gaps in previous studies investigating LDs for the problem of PAL. Overall, lack of standardized methodology and a high risk of bias is a problem when translating results to clinical practice. Low reporting quality further complicates the reproduction of experiments. Despite the limitations of a systematic review to generate conclusive evidence for the best animal model, using negative control groups, disease models and quantifiable outcomes seem important to increase validity and clinical relevance of results. Improvement in animal models are necessary from a scientific, ethical and societal perspective, and we recommend formation of an international panel of experts to further specify guidelines for this field of research.

Supplemental Material

sj-pdf-1-lan-10.1177_00236772231164873 - Supplemental material for Evaluating and developing sealants for the prevention of pulmonary air leakage: A systematic review of animal models

Supplemental material, sj-pdf-1-lan-10.1177_00236772231164873 for Evaluating and developing sealants for the prevention of pulmonary air leakage: A systematic review of animal models by Bob P. Hermans, Steven E.M. Poos, Daniël I.M. van Dort, Jort Evers, Wilson W.L. Li, Erik H.F.M. van der Heijden, Ad F.T.M. Verhagen, Harry van Goor and Richard P.G. ten Broek in Laboratory Animals

Supplemental Material

sj-pdf-2-lan-10.1177_00236772231164873 - Supplemental material for Evaluating and developing sealants for the prevention of pulmonary air leakage: A systematic review of animal models

Supplemental material, sj-pdf-2-lan-10.1177_00236772231164873 for Evaluating and developing sealants for the prevention of pulmonary air leakage: A systematic review of animal models by Bob P. Hermans, Steven E.M. Poos, Daniël I.M. van Dort, Jort Evers, Wilson W.L. Li, Erik H.F.M. van der Heijden, Ad F.T.M. Verhagen, Harry van Goor and Richard P.G. ten Broek in Laboratory Animals

Footnotes

Acknowledgements

The authors would like to thank On Ying Chan from the medical library of the Radboud University Medical Center for help designing the search strategy, and Machteld van Erk for help during the initial design stages of the systematic review.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by GATT-Technologies B.V., Nijmegen, the Netherlands.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: HvG was a scientific advisor for GATT-Technologies B.V until 31 December 2021, but not in relation to lung sealing technology or the current manuscript. The other authors declare that there is no conflict of interest.

Data access statement

All data underlying this article will be made available upon reasonable request.

Collaborators

Chikako Endo (University Medical Center Groningen, Department of Surgery, Groningen, The Netherlands) and Vladimir Khokha (Mozyr City Hospital, Department of Surgery, Mazyr, Belarus).

ORCID iD

Bob P Hermans

Supplemental material

Supplemental material for this article is available online.

References

Attaar

Tam

Nason

KS.

Risk factors for prolonged air leak after pulmonary resection: a systematic review and meta-analysis. Ann Surg 2020; 271: 834–844. DOI: 10.1097/sla.0000000000003560.

Okereke

Murthy

Alster

, et al. Characterization and importance of air leak after lobectomy. Ann Thorac Surg 2005; 79: 1167–1173. DOI: 10.1016/j.athoracsur.2004.08.069.

Yoo

Ghosh

Danker

, et al. Burden of air leak complications in thoracic surgery estimated using a national hospital billing database. Clinicoecon Outcomes Res 2017; 9: 373–383. DOI: 10.2147/ceor.S133830.

Brunelli

Xiume

Al Refai

, et al. Air leaks after lobectomy increase the risk of empyema but not of cardiopulmonary complications: a case-matched analysis. Chest 2006; 130: 1150–1156. DOI: 10.1378/chest.130.4.1150.

Liang

Ivanovic

Gilbert

, et al. Quantifying the incidence and impact of postoperative prolonged alveolar air leak after pulmonary resection. J Thorac Cardiovasc Surg 2013; 145: 948–954. DOI: 10.1016/j.jtcvs.2012.08.044.

Singhal

Ferraris

Bridges

, et al. Management of alveolar air leaks after pulmonary resection. Ann Thorac Surg 2010; 89: 1327–1335. DOI: 10.1016/j.athoracsur.2009.09.020.

Attaar

Luketich

Schuchert

, et al. Prolonged air leak after pulmonary resection increases risk of noncardiac complications, readmission, and delayed hospital discharge: a propensity score-adjusted analysis. Ann Surg 2021; 273: 163–172. DOI: 10.1097/sla.0000000000003191.

Varela

Jimenez

Novoa

, et al. Estimating hospital costs attributable to prolonged air leak in pulmonary lobectomy. Eur J Cardiothorac Surg 2005; 27: 329–333. DOI: 10.1016/j.ejcts.2004.11.005.

Brunelli

Chapman

Pompili

, et al. Ninety-day hospital costs associated with prolonged air leak following lung resection. Interact Cardiovasc Thorac Surg 2020; 31: 507–512. DOI: 10.1093/icvts/ivaa140.

10.

Brunelli

Bölükbas

Falcoz

, et al. Exploring consensus for the optimal sealant use to prevent air leak following lung surgery: a modified Delphi survey from the European Society of Thoracic Surgeons. Eur J Cardiothorac Surg 2020; 59: 1265–1271. DOI: 10.1093/ejcts/ezaa428.

11.

Belda-Sanchis

Serra-Mitjans

Iglesias Sentis

, et al. Surgical sealant for preventing air leaks after pulmonary resections in patients with lung cancer. Cochrane Database Syst Rev 2010: Cd003051. DOI: 10.1002/14651858.CD003051.pub3.

12.

Malapert

Hanna

Pages

, et al. Surgical sealant for the prevention of prolonged air leak after lung resection: meta-analysis. Ann Thorac Surg 2010; 90: 1779–1785. DOI: 10.1016/j.athoracsur.2010.07.033.

13.

McGuire

Yee

Clinical outcomes of polymeric sealant use in pulmonary resection: a systematic review and meta-analysis of randomized controlled trials. J Thorac Dis 2018; 10: S3728–S3739. DOI: 10.21037/jtd.2018.10.48.

14.

Zhou

Lyu

Pang

, et al. Efficiency and safety of TachoSil® in the treatment of postoperative air leakage following pulmonary surgery: a meta-analysis of randomized controlled trials. Jpn J Clin Oncol 2019; 49: 862–869. DOI: 10.1093/jjco/hyz076.

15.

Singhal

Shrager

JB.

Should buttresses and sealants be used to manage pulmonary parenchymal air leaks?

J Thorac Cardiovasc Surg 2010; 140: 1220–1225. DOI: 10.1016/j.jtcvs.2010.06.039.

16.

Macchiarini

Wain

Almy

, et al. Experimental and clinical evaluation of a new synthetic, absorbable sealant to reduce air leaks in thoracic operations. J Thorac Cardiovasc Surg 1999; 117: 751–758. DOI: 10.1016/s0022-5223(99)70296-5.

17.

McCarthy

Trastek

Bell

, et al. The effectiveness of fibrin glue sealant for reducing experimental pulmonary air leak. Ann Thorac Surg 1988; 45: 203–205. DOI: https://dx-doi-org.web.bisu.edu.cn/10.1016/S0003-4975(10)62438-1.

18.

Lewis

Spazierer

Slezak

, et al. Swelling, sealing, and hemostatic ability of a novel biomaterial: a polyethylene glycol-coated collagen pad. J Biomater Appl 2014; 29: 780–788. DOI: 10.1177/0885328214545500.

19.

Annabi

Zhang

Assmann

, et al. Engineering a highly elastic human protein-based sealant for surgical applications. Sci Transl Med 2017; 9: eaai7466 DOI: 10.1126/scitranslmed.aai7466.

20.

Elvin

Vuocolo

Brownlee

, et al. A highly elastic tissue sealant based on photopolymerised gelatin. Biomaterials 2010; 31: 8323–8331. DOI: 10.1016/j.biomaterials.2010.07.032.

21.

Araki

Tao

Nakajima

, et al. Development of new biodegradable hydrogel glue for preventing alveolar air leakage. J Thorac Cardiovasc Surg 2007; 134: 1241–1248. DOI: 10.1016/j.jtcvs.2007.07.020.

22.

Bosmans

Moossdorff

Al-Taher

, et al. International consensus statement regarding the use of animal models for research on anastomoses in the lower gastrointestinal tract. Int J Colorectal Dis 2016; 31: 1021–1030. DOI: 10.1007/s00384-016-2550-5.

23.

van der Worp

Howells

Sena

, et al. Can animal models of disease reliably inform human studies? PLOS Med 2010; 7: e1000245. DOI: 10.1371/journal.pmed.1000245.

24.

Cárdenes

Sembrat

Noda

, et al. Human ex vivo lung perfusion: a novel model to study human lung diseases. Sci Rep 2021; 11: 490. DOI: 10.1038/s41598-020-79434-4.

25.

Horvath

Mueller

, et al. An organosynthetic soft robotic respiratory simulator. APL Bioeng 2020; 4: 026108. DOI: 10.1063/1.5140760.

26.

Klassen

Eckert

Wong

, et al. Ex vivo modeling of perioperative air leaks in porcine lungs. IEEE Trans Biomed Eng 2018; 65: 2827–2836. DOI: 10.1109/tbme.2018.2819625.

27.

Ritskes‐Hoitinga

Leenaars

Avey

, et al. Systematic reviews of preclinical animal studies can make significant contributions to health care and more transparent translational medicine. Cochrane Database of Systematic Reviews 2014; 28: ED000078. DOI: 10.1002/14651858.ED000078.

28.

Hooijmans

Tillema

Leenaars

, et al. Enhancing search efficiency by means of a search filter for finding all studies on animal experimentation in PubMed. Lab Anim 2010; 44: 170–175. DOI: 10.1258/la.2010.009117.

29.

Yauw

Wever

Hoesseini

, et al. Systematic review of experimental studies on intestinal anastomosis. Br J Surg 2015; 102: 726–734. DOI: 10.1002/bjs.9776.

30.

Percie du Sert

Hurst

Ahluwalia

, et al. The ARRIVE guidelines 2.0: updated guidelines for reporting animal research. PLoS Biol 2020; 18: e3000410. DOI: 10.1371/journal.pbio.3000410.

31.

Hooijmans

Rovers

de Vries

, et al. SYRCLE's risk of bias tool for animal studies. BMC Med Res Methodol 2014; 14: 43. DOI: 10.1186/1471-2288-14-43.

32.

Kanzaki

Sekine

Takagi

, et al. Bioartificial pleura using allogenic cell sheet for closing of lung air leakage. JTCVS Tech 2020; 4: 336–340. DOI: https://dx-doi-org.web.bisu.edu.cn/10.1016/j.xjtc.2020.09.024.

33.

Kanzaki

Takagi

Isaka

, et al. Off-the-shelf cell sheets as a pleural substitute for closing visceral pleural injuries. Biopreserv Biobank 2019; 17: 163–170. DOI: 10.1089/bio.2018.0105.

34.

Kanzaki

Yamato

Yang

, et al. Dynamic sealing of lung air leaks by the transplantation of tissue engineered cell sheets. Biomaterials 2007; 28: 4294–4302. DOI: 10.1016/j.biomaterials.2007.06.009.

35.

Balakrishnan

Payanam

Laurent

, et al. Efficacy evaluation of anin situforming tissue adhesive hydrogel as sealant for lung and vascular injury. Biomed Mater 2021; 16: 044106. DOI: 10.1088/1748-605X/abfbbf.

36.

Gika

Kawamura

Izumi

, et al. The short-term efficacy of fibrin glue combined with absorptive sheet material in visceral pleural defect repair. Interact Cardiovasc Thorac Surg 2007; 6: 12–15. DOI: 10.1510/icvts.2006.139436.

37.

Kawai

Suzuki

Naito

, et al. Sealing effect of cross-linked gelatin glue in the rat lung air leak model. Ann Thorac Surg 2016; 102: 282–286. 2016/05/23. DOI: 10.1016/j.athoracsur.2016.02.038.

38.

Ranger

Halpin

Sawhney

, et al. Pneumostasis of experimental air leaks with a new photopolymerized synthetic tissue sealant. Am Surg 1997; 63: 788–795.

39.

Nomori

Horio

Morinaga

, et al. Gelatin-resorcinol-formaldehyde-glutaraldehyde glue for sealing pulmonary air leaks during thoracoscopic operation. Ann Thorac Surg 1999; 67: 212–216. DOI: 10.1016/s0003-4975(98)01184-9.

40.

Servais

Valenzuela

Kienzle

, et al. Functional mechanics of a pectin-based pleural sealant after lung injury. Tissue Eng Part A 2018; 24: 695–702. DOI: 10.1089/ten.tea.2017.0299.

41.

Buyukkale

Citak

Isgorucu

, et al. The effect of sodium hyaluronate-carboxymethyl cellulose membrane in the prevention of parenchymal air leaks: an experimental and manometric study in rats. Tuberk Toraks 2017; 65: 265–270. DOI: 10.5578/tt.9681.

42.

Gigauri

Andrianov

Got'e

, et al. [Hemostasis and hermetic sealing of the parenchyma in operations on the lungs]. Voen Med Zh 1975: 23–25.

43.

Poticha

Macaladad

Lewis

FJ.

The control of air leaks following subsegmental pulmonary resections. Surg Gynecol Obstet 1965; 120: 803–809.

44.

Kausel

Lindskog

GE.

The healing of raw lung surfaces after experimental segmental resection. J Thorac Surg 1955; 29: 197–211.

45.

Wilder

Playforth

Bryant

, et al. The use of plastic adhesive in pulmonary surgery. J Thorac Cardiovasc Surg 1963; 46: 576–588.

46.

Nuchprayoon

Tamayo

Reimann

, et al. The use and tissue reaction of a biologic adhesive in the prevention of air leak following a transection of the lung. Dis Chest 1968; 53: 445–452. DOI: 10.1378/chest.53.4.445.

47.

Attaar

Winger

Luketich

, et al. A clinical prediction model for prolonged air leak after pulmonary resection. J Thorac Cardiovasc Surg 2017; 153: 690–699.e692. DOI: 10.1016/j.jtcvs.2016.10.003.

48.

Moon

Park

Kang

, et al. Significance of the lobe-specific emphysema index to predict prolonged air leak after anatomical segmentectomy. PLoS One 2019; 14: e0224519. DOI: 10.1371/journal.pone.0224519.

49.

DeCamp

Blackstone

Naunheim

, et al. Patient and surgical factors influencing air leak after lung volume reduction surgery: lessons learned from the National Emphysema Treatment Trial. Ann Thorac Surg 2006; 82: 197–206; discussion 206–197. DOI: 10.1016/j.athoracsur.2006.02.050.

50.

Nam

Mooney

. Polymeric tissue adhesives. Chem Rev 2021; 121: 11336–11384. DOI: 10.1021/acs.chemrev.0c00798.

51.

Rivera

Bernard

Falcoz

, et al. Characterization and prediction of prolonged air leak after pulmonary resection: a nationwide study setting up the index of prolonged air leak. Ann Thorac Surg 2011; 92: 1062–1068; discussion 1068. DOI: 10.1016/j.athoracsur.2011.04.033.

52.

Mentzer

Tsuda

Loring

SH.

Pleural mechanics and the pathophysiology of air leaks. J Thorac Cardiovasc Surg 2018; 155: 2182–2189. DOI: 10.1016/j.jtcvs.2017.12.062.

53.

Pompili

Miserocchi

Air leak after lung resection: pathophysiology and patients' implications. J Thorac Dis 2016; 8: S46–S54. DOI: 10.3978/j.issn.2072-1439.2015.11.08.

54.

Zaraca

Brunelli

Pipitone

, et al. A Delphi consensus report from the “Prolonged Air Leak: A Survey” study group on prevention and management of postoperative air leaks after minimally invasive anatomical resections. Eur J Cardiothorac Surg 2022; 62: ezac211 DOI: 10.1093/ejcts/ezac211.

55.

Brunelli

Salati

Pompili

, et al. Intraoperative air leak measured after lobectomy is associated with postoperative duration of air leak. Eur J Cardiothorac Surg 2017; 52: 963–968. DOI: 10.1093/ejcts/ezx105.

56.

Kim

Lee

Ryu

, et al. Intraoperative ventilatory leak predicts prolonged air leak after lung resection: a retrospective observational study. PLoS One 2017; 12: e0187598. DOI: 10.1371/journal.pone.0187598.

57.

Charan

Kantharia

ND.

How to calculate sample size in animal studies?

J Pharmacol Pharmacother 2013; 4: 303–306. DOI: 10.4103/0976-500x.119726.

58.

Rooney

Mehl

Monnet

Intercostal thoracotomy closure: transcostal sutures as a less painful alternative to circumcostal suture placement. Vet Surg 2004; 33: 209–213. DOI: 10.1111/j.1532-950X.2004.04031.x.

59.

de Vries

Wever

Avey

, et al. The usefulness of systematic reviews of animal experiments for the design of preclinical and clinical studies. ILAR J 2014; 55: 427–437. DOI: 10.1093/ilar/ilu043.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB

0.35 MB