Pumpless extracorporeal interventional lung assist in patients with acute respiratory distress syndrome: a prospective pilot study

Pumpless extracorporeal interventional lung assist (iLA; Novalung, Talheim, Germany) has been described in patients with life-threatening forms of respiratory failure or acute respiratory distress syndrome (ARDS) suffering from persistent hypoxaemia and/or hypercapnia, unresponsive to conventional therapy [1, 2]. The iLA system is characterised by a novel membrane gas exchange device with optimised blood flow integrated into an arteriovenous heparin-coated bypass, established by cannulation of the femoral artery and vein. A passive shunt flow generated by the patient's blood pressure gradient through the gas exchange device allows effective carbon dioxide extraction and moderate improvement in arterial oxygenation [3].

We previously examined iLA implementation in a heterogeneous group of 90 severe critically ill patients at our institution and reported a hospital mortality of 59% with a complication rate of 24.4% [4]. Specifically, the patient groups with the highest mortality were identified as patients with cancer, a high demand for vasopressors, advanced age, morbid obesity or those requiring long duration of mechanical ventilation. The most frequent complications reported were related to the arterial cannulation of the femoral vessels, resulting in distal limb ischaemia, compartimental syndrome [4] or bleeding from the cannulation sites requiring surgical intervention [1, 5, 6].

To date, solid evidence-based outcome data for iLA does not exist, in part, because of inherent patient heterogeneity and the rescue nature associated with the therapy, thereby leaving most iLA protocols to be determined individually 'per case'. An experience-based algorithm-guided approach to implement arteriovenous iLA in the treatment of ARDS has recently been published, setting new criteria for care [7]. The unmet need of clinicians for systematically attained, well-structured studies to reliably define indications for iLA sets a high value on publications of this nature and underscores the importance of establishing proven algorithms for ARDS therapy.

The present prospective pilot study defines an algorithm for cannulation and implementation of iLA while refining inclusion and exclusion criteria because we found evidence that arterial insertion of cannulae with lower diameter might reduce the rate of ischaemic complications. In contrast, the hypothesis that timely initiation of iLA in an early phase of severe respiratory failure, and a consequent tidal volume reduction during iLA therapy will positively influence outcome, will be determined in future randomised controlled trials.

Initiation of iLA resulted in a significant improvement in arterial oxygenation and a marked removal in arterial carbon dioxide within two hours allowing a rapid reduction in F_iO₂, minute ventilation and inspiratory plateau pressure (Table 2). Furthermore, we were able to set lung protective tidal volume consequently to a level below 6 ml/kg IBW without provoking severe hypercapnia and/or acidosis. Following our PEEP/FiO₂ trial, we found no changes in the PEEP level within 24 hours after starting iLA. Insertion of iLA did not induce haemodynamic instability although mean arterial pressure or the amount of continuous infusion of noradrenaline remained unchanged, or showed a tendency towards stabilisation. Sequential organ failure assessment (SOFA)-score did not change significantly after 24 hours following initiation of iLA.

Twenty-six of 51 patients (50.9%) survived ARDS, non-survivors had a significant higher age, although the severity of disease (SOFA-score) and the severity of lung injury (Lung Injury Score) were not different between survivors and non-survivors (Table 3).

The frequency of complications is reviewed in Table 4. Transient episodes of lower limb ischaemia after arterial cannulation occurred in three patients. This resulted in removal of the cannula, resulting in normalisation of distal perfusion. Surgical intervention was mandatory in one patient who developed a compartment syndrome. From those four ischaemic complications, three patients had been cannulated with a 17 Fr cannula for the arterial approach. Other complications were rarely seen (cannula thrombosis, bleeding). In a total of six patients (11.9%) adverse events were observed. No adverse event had an effect on outcome.

The main results of our prospective case series are: a change in the indication spectrum resulted in a trend toward an increased survival rate compared with the retrospective comparator study [4]; and a significant reduction in the incidence of adverse events, especially ischaemic complications. In our patients, iLA enabled a safe application of lung protective tidal volume (V_T = 6 ml/kg IBW) and in some patients even less than 6 ml/kg without provoking severe acidosis. The combination of very low V_T with high PEEP allowed the limitation of plateau pressure at 30 cmH₂O or lower and, thus the avoidance of barotrauma. The use of smaller cannulae and an improved cannulation technique (careful assessment of the vessels by ultrasound, cowork-insertion by two experienced physicians) may facilitate the reported improvements. With smaller cannulae (arterial 17 Fr or lower), a sufficient blood flow of 1.0 to 1.5 L/minute allowed for adequate carbon dioxide-removal within the circuit [12]. Although the support of a strict lung protective ventilation [13] is our main goal for the use of iLA in ARDS patients, the insertion of smaller cannulae is sufficient to reach such a goal and minimises adverse ischaemic events. Improvements in cannulae allowed the use of shorter (9 cm versus 14 cm) and thinner (13 versus 15 Fr) cannulae for arterial cannulation. It is conceivable that further evolution of this technology will contribute to greater improvements in the risk-benefit ratio for the use of arteriovenous iLA.

Over the past three decades, an intensive scientific debate on the effectiveness and possible harm of extracorporeal lung assist systems was stimulated by clinical investigations studying ECMO or extracorporeal carbon dioxide removal in rescue situations (life-threatening hypoxaemia/hypercapnia) of patients with ARDS [14‐17]. In small prospective randomised investigations, no clear survival benefit was demonstrated using ECMO in comparison to 'conventional' treatment, and the complication rate was shown to be high (more than 50%) [14]. From today's point of view there were two major limitations regarding the results and the interpretation of previous studies. Specifically, patients were ventilated in the mode of the 'Pre-ARDSNetwork-era' with high tidal volumes and relatively low PEEP, thus a harmful potential of ventilator-induced lung injury was postulated; and the 'historic' use of roller pumps might have aggravated the high complication rate (demand for elevated doses of heparin, induction of haemolysis) and in the past, the technique of miniaturised centrifugal pumps has been advocated [18‐20].

A retrospective analysis of a new iLA system using an arteriovenous shunt and a membrane lung, characterised by an extremely-low flow resistance, demonstrated effective carbon dioxide removal and a moderate oxygenation improvement in severe ARDS, but no survival benefit in life-threatening hypoxaemia/hypercapnia was observed [4]. This was potentially because of the lack of a clear algorithm and indication strategy. Since 2004, we have worked to change the use of iLA in terms of defining indications, refining insertion technique and optimising the management of possible complications. In contrast to a previously described clinical concept aimed at 'rescue' in life-threatening gas exchange limitation, the new algorithm introduced in this report stipulates the insertion of iLA mainly for extracorporeal carbon dioxide-removal purposes facilitating lung-protective ventilation. In our view, the most important aspect of the presently described algorithm is the indication strategy. In summary, we withdrew the iLA-system from life-threatening 'rescue-situations' towards the support of lung-protective ventilation in acute lung injury or early ARDS on the threshold of becoming 'established' ARDS [10]. In this concept, patients with a critically impaired pulmonary gas exchange remain candidates for ECMO.

A comparison of cannula diameter, iLA effects on gas exchange and complications between the present study and our earlier work [4] demonstrated that the improved methodology described here led to a significant reduction in the rate and nature of complications. In the present prospective pilot study, six patients (11.8%) had complications because of iLA (transient ischaemia of a lower limb in three patients), while in our recent retrospective analysis the incidence of complications was 24.4%, and serious ischaemic complications occurred in nine patients (10%, p < 0.05). Furthermore, there is a trend toward a decrease in mortality in our prospective cohort (49.1%) compared with the retrospectively analysed patients suffering from ARDS (58.9%).

Nevertheless, our present analysis has some limitations. Having been conducted in a single-centre, the study might be subject to bias. Furthermore, the data stem from an expert team of intensivists and perfusionists, having a long 'learning curve' experience (> 150 applications) with the iLA system. Taking this into consideration, we present – in contrast to our previous analysis – clinical data resulting from a strict algorithm with defined indications (and contraindications).

Importantly. however, was also the identification of patient groups that do not presently receive benefit from our iLA algorithm. Specifically, in haemodynamically unstable patients requiring high doses of vasopressors (noradrenaline 0.4 μg/kg/minute or higher) or in patients with severe hypoxaemic ARDS, a pump-driven ECMO is still the rescue measure of choice.

Our new algorithm specifies indications for iLA differing from ECMO that underscore the respective differences in therapy concept. Although ECMO, characterised by high blood flow, resembles an 'artificial lung' by producing significant exchange of carbon dioxide and oxygen, iLA with its low blood flow provides impressive carbon dioxide elimination with a modest oxygenation improvement. Consequently iLA could be used as an adjunct to mechanical ventilation affording optimised lung-protective ventilation strategies, with the objective of giving the lungs time to heal [21].

Our data demonstrate that iLA can be an important tool enabling advanced lung-protective ventilation in patients suffering from ARDS. The use of an indication algorithm for iLA in early ARDS, combined with a refined application technique was associated with efficient carbon dioxide removal and a reduction in the incidence of adverse events, especially ischaemia complications. With the ongoing technical evolution of smaller cannulae, more efficient gas exchange membranes and easy system handling, we hypothesise that iLA could serve as an extracorporeal assist to support respirator ventilation by enabling low tidal volume and reduced inspiratory plateau pressure as an important tool in ARDS management. This hypothesis is currently being tested by a prospective multicentre randomised trial (ClinicalTrials NCT 00538928).