Proof-of-principle of a lung sealant based on functionalized polyoxazolines: experiments in an ovine acute aerostasis model

Abstract

OBJECTIVES

More effective lung sealants are needed to prevent prolonged pulmonary air leakage (AL). Polyoxazoline-impregnated gelatin patch (N-hydroxysuccinimide ester functionalized poly(2-oxazoline)s; NHS-POx) was promising for lung sealing ex vivo. The aim of this study is to confirm sealing effectiveness in an in vivo model of lung injury.

METHODS

An acute aerostasis model was used in healthy adult female sheep, involving bilateral thoracotomy, amputation lesions (bronchioles Ø > 1.5 mm), sealant application, digital chest tube for monitoring AL, spontaneous ventilation, obduction and bursting pressure measurement. Two experiments were performed: (i) 3 sheep with 2 lesions per lung (N = 4 NHS-POx double-layer, N = 4 NHS-POx single-layer, N = 4 untreated) and (ii) 3 with 1 lesion per lung (N = 3 NHS-POx single-layer, N = 3 untreated). In pooled linear regression, AL was analysed per lung (N = 7 NHS-POx, N = 5 untreated) and bursting pressure per lesion (N = 11 NHS-POx, N = 7 untreated).

RESULTS

Baseline AL was similar between groups (mean 1.38–1.47 l/min, P = 0.90). NHS-POx achieved sealing in 1 attempt in 8/11 (72.7%) and in 10/11 (90.9%) in >1 attempt. Application failures were only observed on triangular lesions requiring 3 folds around the lung. No influences of methodological variation between experiments was detected in linear regression (P > 0.9). AL over initial 3 h of drainage was significantly reduced for NHS-POx [median: 7 ml/min, length of interquartile range: 333 ml/min] versus untreated lesions (367 ml/min, length of interquartile range: 680 ml/min, P = 0.036). Bursting pressure was higher for NHS-POx (mean: 33, SD: 16 cmH₂O) versus untreated lesions (mean: 19, SD: 15 cmH₂O, P = 0.081).

CONCLUSIONS

NHS-POx was effective for reducing early AL, and a trend was seen for improvement of bursting strength of the covered defect. Results were affected by application characteristics and lesion geometry.

Prolonged pulmonary air leakage (pPAL), an air leak that persists for more than 5 days, occurs in 5–30% of patients after lung resection [1]. pPAL is associated with an increased risk of postoperative complications (i.e. empyema), reinterventions, re-admissions and mortality [2–5].

Graphical abstract

INTRODUCTION

Prolonged pulmonary air leakage (pPAL), an air leak that persists for more than 5 days, occurs in 5–30% of patients after lung resection [1]. pPAL is associated with an increased risk of postoperative complications (i.e. empyema), reinterventions, re-admissions and mortality [2–5]. On average, the hospital stay is extended by 4 days, and hospital costs are 50% higher compared to patients without pPAL [6]. Lung sealants to prevent pPAL are advocated; however, current sealants have not been satisfactorily proven to be effective on clinically relevant outcomes, such as length of hospital stay, despite good preclinical results [7–9]. An unmet clinical need exists for a better product that is easily applicable, capable of hermetic lung sealing in a wet environment and sufficiently compliant for lung expansion [8].

Polyoxazolines are a novel group of polymers that hold promise as a lung sealant due to their strong adhesive and cohesive characteristics [10–13]. A specific formulation containing amine-functionalized and N-hydroxysuccinimide ester functionalized poly(2-oxazoline)s (NHS-POx, GATT Technologies BV, Nijmegen, The Netherlands) was a highly effective haemostat when impregnated in a porcine gelatin carrier [10, 11, 13]. NHS-POx reacts with proteins on tissue and blood within the patch to form covalent bonds, creating a strong sealing hydrogel within minutes [10]. We have previously demonstrated lung-sealing effectiveness with these patches ex vivo, showing superior aerostatic efficacy compared to sprays based on polyethylene glycol (Progel^®, Coseal^®) and collagen patches with polyethylene glycol (Hemopatch^®) or fibrin/thrombin coating (TachoSil^®) [14].

We aim to further confirm the effectiveness of the NHS-POx patch as an aerostatic lung sealant using a clinically relevant sheep lung injury model.

METHODS

Ethical statement

Experiments were performed under a project license (AVD10300202114869) granted by national authorities (4-10-2021) after review by an ethics board. Protocols were approved by the Animal Welfare Body and registered at our institute (2021-0012-002).

Study design

The in vivo study was performed in 2 phases, comparing the NHS-POx patch to a control group (Fig. 1). Lung lesions resulting in clinically relevant pulmonary air leakage were made in sheep, treated with NHS-POx and observed during spontaneous ventilation, followed by measuring air leakage (AL) and bursting pressure (BP) [15]. The 1st experiment was performed with 3 animals (E1–E3) with 4 lesions per animal, 2 lesions in each lung, based on a sample size calculation. An additional 3 animals (E4–E6) with 2 lesions per animal, 1 lesion in each lung, were included in the 2nd experiment. Randomization was performed per lung, ensuring: 1 group per lung, no 2 same groups per animal and each group once to each lung. Allocation sequence was concealed until application.

Figure 1:

Study flowchart.

After the 1st experiment, we observed that the right middle lobe/left upper lobe lesions were more triangular in shape, causing application difficulties. These lesions are not clinically representative in shape and location compared to the lesions that generally result from lung surgery for which a sealant would be used. Therefore, only right lower lobe/left lower lobe lesions were included in the 2nd experiment. Additionally, 2 drains were used per lung in the 2nd experiment to ensure adequate drainage during the contralateral thoracotomy, and BP was measured in situ to prevent excessive manipulation and potential blood contact of lesions during obduction.

Anaesthesia and surgery

Adult female sheep (Ovis aries) were housed with buddy sheep at least 1 day before surgery, and fed ad libitum. Pre-medication (ketamine and midazolam) was administered before induction of anaesthesia (remifentanil, propofol, carprofen). Anaesthesia was maintained using remifentanil, propofol and isoflurane (no isoflurane with active AL). An endo-tracheal tube, urinary catheter and blood pressure catheter (femoral artery) were placed.

Mechanical ventilation settings were adjusted to match ventilation and oxygenation requirements, maintaining pressure <25 cmH₂O. Sequentially, right and left lateral thoracotomies were made and closed in the 5th intercostal space, starting with the right lung. An intercostal block was placed on 3 levels with lidocaine/bupivacaine. After lesion creation and sealing, a chest tube was positioned apically in the thorax. In the 2nd experiment, an additional chest tube was placed ventrally in the thorax. Incisions were closed airtight in layers, and drainage was started after incision closure.

Spontaneous ventilation was observed for 3–4 h under ketamine, midazolam and buprenorphine anaesthesia. Vital signs and arterial blood gas were measured throughout the experiment. After observation, euthanasia was performed using pentobarbital. Obduction was performed, and observations were noted. The lungs were further used for BP measurements, and samples were taken for histological processing.

Defect creation and sealing

Strips of tissue at a maximum width of 5 cm were removed from the right middle lobe, left upper lobe, right lower lobe and left lower lobe in steps of 0.5–1 cm, until bronchioles of 1.5 mm were encountered (Video 1, Fig. 1). Haemostasis was achieved using diathermy and compression. Fibrin plugs were removed from the bronchiole to ensure pulmonary air leakage. No further treatment was performed in the control group. In the NHS-POx group, the lung was primed using phosphate-buffered saline (PBS), which improves crosslinking and adhesive strength. The patch was applied with >0.5 cm on healthy pleura and pressure was applied (2 min, gauzes with PBS/saline). For the double layer, a 2nd patch was applied, irrigating with 20 cc PBS/saline solution in between the patches, before applying pressure. In case of application failure, a new patch was applied (application fluids: Table 1).

Table 1:

Aerostatic findings

ID	Group		AL (ml/min)	Lobe (s)	Sealant	Fluids	Acute sealing	BP (cmH2O)
Experiment 1
E1	R	Seal.	3	RML	NHS-POx/single	PBS	Yes, 2nd attempt	50
				RLL	NHS-POx/single	PBS	Yes	30
	L	Seal.	2	LUL	NHS-POx/double	PBS	Yes	35
				LLL	NHS-POx/double	PBS	Yes	20
E2	R	Cont.	886	RML	Untreated	N/A	N/A	9 b
				RLL	Untreated	N/A	N/A	4 b
	L	Seal.	336	LUL	NHS-POx/single	PBS/saline	No, 4th attempt	20
				LLL	NHS-POx/single	Salinec	Yes	40
E3	R	Seal.	14	RML	NHS-POx/double	PBS/saline	Yes, 2nd attempt	40
				RLL	NHS-POx/double	PBS/saline	Yes	25
	L	Cont.	913	LUL	Untreated	N/A	N/A	15
				LLL	Untreated	N/A	N/A	45
Experiment 2
E4	R	Cont.	267	RLL	Untreated	N/A	N/A	29 b
	L	Seal.	7	LLL	NHS-POx/double	Salinea	Yes	60
E5	R	Seal.	6	RLL	NHS-POx/double	Salinea	Yes	40
	L	Cont.	173	LLL	Untreated	N/A	N/A	23 b
E6	R	Seal.	1202	RLL	NHS-POx/double	Salinea	Yes	4 b
	L	Cont.	367	LLL	Untreated	N/A	N/A	6 b

^aPBS priming.

^bMinimal leaking pressure protocol.

AL: average air leak first 3 h of drainage; BP: bursting pressure; L: left; LLL: left lower lobe; LUL: left upper lobe; PBS: phosphate-buffered saline; R: right; RML: right middle lobe; RLL: right lower lobe; N/A = not applicable

Outcome measures

Baseline AL was measured by registering tidal volumes for 15 breaths [respiratory rate 15/min, positive end expiratory pressure (PEEP) 5 cmH₂O, plateau ventilator pressure 15–20 cmH₂O]. Baseline minimal leaking pressure was determined by adjusting PEEP (steps of 1 cmH₂O, starting at 10 cmH₂O). Sealing success was defined by the absence of bubbles after irrigation with saline directly after application. Postoperative AL was measured using a digital drainage system (Thopaz^®, Medela, Baar, Switzerland). BP (pressure at 1st visual bubbles) is determined post-mortem, by increasing ventilation pressure with 5 cmH₂O every 60 s [respiratory rate 12/min, PEEP 5 cmH₂O, inspiratory expiratory ratio (I:E) 1:2, pressure control above PEEP (Pc) 15 cmH₂O] [14]. For control lesions or sealed lesions with persistent leakage, minimal leaking pressure was determined and analysed as a BP value. Failure mechanism was noted as adhesive (leakage between the sealant and lung) or cohesive (leakage within the sealant). Histology samples were stored in 4% formaldehyde, embedded in paraffin, sliced in (4 μm), stained with haematoxylin–eosin and analysed for signs of acute adverse inflammatory response.

Data analysis

Postoperative AL data were imported using ThopEasy+ software (Medela, Baar, Switzerland), giving mean AL values (ml/min) every 10 min. Mean postoperative AL was calculated over the first 3 h after start of drainage. This asynchronous interval was chosen to facilitate interpretation in the context of time-dependent intrinsic sealing [15]. Intraoperative baseline AL (ml/min) based on the mechanical ventilator was calculated as: $AL=\frac{{\sum }_{k=1}^5({\mathit{TVi}}_k-{\mathit{TVe}}_k)}{5}\times RR$ [16]. This was corrected for AL measured before lesion creation as: ${AL}_{\mathit{corrected}}={AL}_{\mathit{lesion}}-{AL}_{\mathit{baseline}}$.

As the sample size was small and possible differences are detectable with more power on a continuous scale, we used linear regression to estimate differences in mean postoperative AL and BP between sealing and non-sealing while correcting for experiment heterogeneity (using the intervention group and experiment number as predictors). NHS-POx samples were pooled and compared to control lesions (Fig. 1). AL was log-transformed and BP was square-root transformed to ensure normal distribution of residuals. Significance was considered if P < 0.05 for predictors in the analysis (see Supplementary Material, Section ‘Linear regression’ for details). Baseline values were compared between groups using a two-sided independent samples T-test (continuous data) or Fischer’s exact test (ordinal data) with a significance level of 0.05. IBM SPSS Statistics 27 (Armonk, New York; IBM Corp) was used for statistical testing.

RESULTS

Similar baseline AL (mean 1.38–1.47 l/min, P = 0.90, all Macchiarini grade 3), minimal leaking pressure (5 cmH₂O, P = 0.85), bleeding (mean grade 3.3–3.8, P = 0.27) and other baseline characteristics were observed (Supplementary Material, Table S1) [17, 18]. There are no evident inter-sheep correlations in baseline leakage characteristics (Supplementary Material, Fig. S1).

Sealing using NHS-POx was achieved in 8/11 (72.7%) cases in 1 attempt (P = 0.004) and in 10/11 (90.9%) cases in >1 attempt (P < 0.001). Failures occurred only on lesions with a triangular shape (right middle lobe/left upper lobe lesions, P = 0.024, Fig. 2, Table 1). Saline seemed to cause less unwanted sticking of patch to gauzes compared to PBS. Compression atelectasis was observed after application (Fig. 4).

Figure 2:

(A) Middle/upper lobe (RML/LULca) lesions are more triangular (B) and lower lobe (RLL/LLL) lesions are more oval (C). Reused/adapted with permission from AME Publishing Company [15]. LLL: left lower lobe; LUL: left upper lobe; RLL: right lower lobe; RML: right middle lobe.

Figure 4:

(A) Patch on middle/upper lobe. (B) Mode of failure double patch. Dome formation with cohesive tear through the dome (half of dome cut away). Arrow pointing at bronchiole. (C) Double patch, showing atelectasis and folding of lung parenchyma (white arrow). (D) Mode of failure of patch in (C), showing cohesive failure due to lung expansion (dotted lines).

Sealing with NHS-POx was associated with a lower postoperative AL (median: 7 ml/min, length of interquartile range: 333 ml/min) compared to the control group (median: 367 ml/min, length of interquartile range: 680 ml/min, P = 0.036) in regression analysis. Experiment heterogeneity was not significant (P = 0.936) (Fig. 3). Control lungs showed intrinsic sealing (AL <10 ml/min) in 3/5 (60%) cases, within a mean of 2:57 h (SD: 1:45 h) after start of drainage.

Figure 3:

(A) Air leak over first 3 h of drainage. (B) Bursting pressure after obduction.

BP was higher for the NHS-POx group (mean: 33, SD: 16 cmH₂O) compared to the control group (mean: 19, SD: 15 cmH₂O), but not significant in regression analysis (P = 0.081) (Fig. 3). Experiment heterogeneity was not significant (P = 0.985). Mode of failure was cohesive in 10/11 (90.9%) NHS-POx sealed cases (Examples Fig. 4B and D).

At sacrifice, migration of a small piece was seen in 3 samples (E3R and E4L). Some swelling of the patch was observed and appeared lower in the 2nd experiment due to the ventral chest tube. The patch appeared less adhesive and cohesive compared to right after application. Morphological analysis revealed a response consistent with the created lung injury (collapsed parenchyma, fibrin depositions, neutrophile infiltration). No major acute inflammatory reactions were observed (Fig. 5).

Figure 5:

(A) Example of pleural interface, showing tight adhesion and normal alveoli. (B) Detail of (A), showing influx of neutrophilic granulocytes (arrows).

DISCUSSION

We have investigated novel NHS-POx-coated gelatine patches, demonstrating proof-of-principle for lung sealing in sheep mimicking the clinical situation. In the majority of cases, clinically severe ALs, on average measuring >400 ml/min, could be sealed with 1 application [19]. Immediate postoperative AL was significantly reduced compared to non-treated lung lesions, and the bursting strength of the covered lesions tended to be better than the controls, indicative of treatment effectiveness.

Only 1 adhesive failure was noted, confirming adequate adhesion to the tissue [8]. In contrast, the currently most studied lung-sealing patch (TachoSil^®) showed mainly adhesive failures in preclinical studies [14, 20, 21]. For lung sealing, an optimal balance is required between adhesive and cohesive forces, but a cohesive failure mechanism may be advantageous. In our previous study, we noted that patches with an adhesive failure mechanism generally lost all sealing potential upon bursting, whereas samples with cohesive failure mechanisms showed gradual loss of sealing potential due to strong adhesion to the leaking surface [14].

Issues regarding surgical application need to be addressed. First, adhesion of the gauzes to the patch were seen, and the application fluids were adjusted (Table 1). Second, the double layer showed a specific mode of failure. As irrigation is required between the patches for cohesion, no immediate pressure is applied, allowing an air pocket to form, which can cause dome formation and cohesive tearing [14].

Application failure occurrence differed significantly based on lesion geometry, and these were only seen on triangular lesions, which is an unconventional geometry in clinical practice [20]. Comparison of sealant efficacy on various geometries is not well studied, but lesions requiring several folds around edges seem less suited for patches. Conceptually, spray sealants with high viscosity might be better in such cases. The sealing efficacy of the patch might be higher on other geometries, which were not studied, such as superficial lesions and dissected fissures [14].

A higher BP was observed for the NHS-POx samples, without reaching statistical significance. However, 7/11 (64%) of sealed samples withstood pressures >30 cmH₂O, which is clinically relevant as most anaesthesiologists attempt to limit ventilatory pressure to a maximum of 30 cmH₂O [22]. During lung explantation for ex-situ measurements, blood contact of the control lesions was observed, which may have increased BP, while the manipulations might have damaged the sealants, lowering the BP. Possibly, a significant difference would have been demonstrated, when all measurements were performed in situ.

Comparison to relevant literature is difficult due to many studies lacking a standardized model of pulmonary air leakage, appropriate control groups and quantitative outcome measures [23, 24]. Intrinsic sealing mechanisms in healthy animals may invalidate results if no control groups are used [15]. Because AL is mainly due to alveolar leakage in patients with emphysema, the translational value from healthy animal lungs remains unclear, and lung-sealing results may be overestimated due to a lower crosslink density in emphysema [1, 15, 25]. However, due to the use of large bronchiole lesions, the current model likely poses a more challenging case than emphysema, making an overestimation of sealing efficacy unlikely. The validity of this model has been discussed extensively in our previous work [15].

No comparison to similar devices was performed, but superior BP was previously measured ex vivo [14]. Further in vivo comparison to gold-standard products should be performed for benchmarking. Clinically, the best studied patch is the fibrinogen–thrombin-coated collagen patch (TachoSil^®) [21]. Gel/spray sealants, such as the novel human serum albumin/polyethylene glycol spray (Progel^®) or the polyethylene glycol spray (Coseal^®) appear less suited for comparison to a patch, due to the aberrant mode of application. The polyethylene glycol-coated collagen patch (Hemopatch^®) could also be a good contender for benchmarking due to similar composition and mode of application (trial ID: NCT03450265).

We showed that 5/7 (71.4%) patches had <20 ml/min postoperative AL. For comparison, several studies with negative control groups and AL measurements should be pointed out, as they show comparable results. Ranger et al. used similar lesions to test a photopolymerizable gel, which remained aerostatic in 80% of cases over 24 h [23]. Kjaergard et al. measured AL for 2 h over bilateral upper lobe wedge resection, demonstrating an average 92% AL reduction with an autologous fibrin sealant [9]. McCarthy et al. used large parenchymal lesions, which were either left untreated or sealed with fibrin glue, demonstrating an average reduction of 19.8% vs 80.8% in AL from baseline [26].

With regard to safety, no major acute adverse inflammatory or allergic reactions were observed. Other studies have shown biocompatibility and biodegradability within 6 weeks for intra-abdominal use, and the polymers are renally excreted [11–13]. However, the present experiment has limited value with regard to demonstrating safety for intrathoracic use. Organ-specific safety considerations should be further scrutinized in longer-term experiments, relating to biodegradation as a function of underlying tissue healing, material migration, foreign body response and impact on adhesion formation, lung expansion, atelectasis, infection and obstruction of chest tubes.

Macroscopically, swelling appeared to impact the adhesive–cohesive matrix integrity at sacrifice. This might pose a risk in the postoperative period for re-leakage. Swelling occurs due to the hydrophilicity designed to enhance blood uptake for haemostatic purposes, but will be mainly due to water uptake in primary air leaks [10]. This increases polymer chain separation, which weakens secondary intermolecular forces, and enhances hydrolysis [12, 27, 28]. Specifications with regard to degradation speed as a function of underlying tissue healing are unknown, and should be further investigated. However, re-leakage may be challenging to investigate in animal studies, since there are no valid long-term pPAL models currently, so histological aspects of wound healing and biodegradation may need to be used as a surrogate.

These experiments consist of both model development and proof-of-principle of a new lung sealant, forming the basis for further investigations. Methodological variation between the experiments and a small sample size are inherent limitations. By use of linear regression and pooling of NHS-POx samples, we were able to account for this variation. The main limitation of this approach is the small sample size and presence of outliers (see Supplementary Material Section ‘Linear regression’), so the results should be interpreted with caution. Ideally, this analysis should also correct for biological variability based on animal and lung side effects, but the sample size is considered to be too small for this to be valid.

Postoperative AL data were analysed per lung over the first 3 h of drainage to allow for meaningful interpretation of AL in the context of rapid intrinsic sealing mechanisms. Consequently, lesions were exposed to different ventilation mechanics. Based on additional analysis of our previous model validation study, no significant effects were found between left and right lungs (Supplementary Material, Table S2) [15].

Before clinical applications, further study is required. The single-patch should be improved to replace the double patch, and the application method should be optimized to minimize sticking to gauzes. Comparison to currently available similar devices is required for benchmarking. Due to the limited sample size and heterogeneity in the current study, larger studies are needed for validation. Finally, in vivo biodegradability, safety and healing of underlying tissue should be studied in longer-term survival models, before these findings can be translated to clinical practice.

CONCLUSIONS

Based on this proof-of-principle study, NHS-POx technology seems a potent lung sealant, based on effective reduction of severe ALs in a subacute model. The mode of application, including application fluids and effects of lesion geometry, needs to be further studied to meet the clinical demands of a lung sealant. Further research is required for longer-term aerostatic efficacy and re-leakage, benchmarking, biodegradability and safety.

Glossary

ABBREVIATIONS

AL

Air leakage

NHS-POx

N-hydroxysuccinimide ester functionalized poly(2-oxazoline)s

PBS

Phosphate buffered saline

PEEP

Positive end expiratory pressure

pPAL

Prolonged pulmonary air leakage

SUPPLEMENTARY MATERIAL

Supplementary material is available at ICVTS online.

FUNDING

This work was supported by GATT-Technologies B.V. (Nijmegen, The Netherlands).

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request.

Author contributions

Bob P. Hermans: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Visualization; Writing—original draft; Writing—review and editing. Richard P.G. Ten Broek: Formal analysis; Supervision; Writing—original draft; Writing—review and editing. Wilson W.L. Li: Conceptualization; Formal analysis; Methodology; Supervision; Writing—review and editing. Edwin A. Roozen: Conceptualization; Investigation; Methodology; Writing—review and editing. Shoko Vos: Formal analysis; Writing—review and editing. Erik H.F.M. Van Der Heijden: Formal analysis; Supervision; Writing—review and editing. Harry Van Goor: Conceptualization; Formal analysis; Methodology; Supervision; Writing—review and editing. Ad F.T.M. Verhagen: Conceptualization; Formal analysis; Funding acquisition; Methodology; Supervision; Writing—review and editing.

Reviewer information

Interactive CardioVascular and Thoracic Surgery thanks Lucio Cagini, Gaurav Sharma and the other anonymous reviewers for their contribution to the peer review process of this article.

 

Presented at the 37th EACTS Annual Meeting in Vienna (Austria) on 7 October 2023.