Targeting the diaphragm during weaning from mechanical ventilation

Although mechanical ventilation (MV) is an essential supportive therapy in patients with acute respiratory failure (1), it is associated with well-recognized complications. These include ventilator-associated pneumonia, ventilator-induced lung injury, and ventilator-induced diaphragmatic dysfunction (2). The risk of these complications increases with longer MV duration, and their occurrence may further prolong ventilatory support. This creates a self-perpetuating cycle that impairs liberation from MV (3).

MV may contribute to the development of ventilator-induced diaphragmatic dysfunction through several mechanisms (4). Among these, diaphragmatic disuse atrophy resulting from controlled or excessive ventilatory assistance is considered the predominant pathway. Clinical studies have shown that patients receiving full ventilatory support or excessive assistance develop diaphragmatic muscle fiber atrophy and injury, mediated by oxidative stress, inflammation, and enhanced proteolysis (5). These structural alterations are associated with reduced diaphragmatic force-generating capacity (6). Conversely, inadequate ventilator settings that impose excessive inspiratory effort may also lead to diaphragmatic injury, particularly when sustained over time, and may adversely affect the weaning process (7). Patient–ventilator asynchrony may result in eccentric muscle injury, especially with diaphragmatic contraction during expiration (8). High levels of positive end-expiratory pressure (PEEP) or dynamic hyperinflation may shorten diaphragmatic fiber length. This leads to longitudinal atrophy and impaired force generation (9). Other factors in critically ill patients—such as sepsis, chronic heart failure, and chronic obstructive pulmonary disease—can further exacerbate diaphragmatic dysfunction. Exposure to corticosteroids, aminoglycosides, neuromuscular blocking agents, and inadequate protein delivery may also worsen diaphragmatic dysfunction (10,11).

Diaphragmatic weakness from ventilator-induced diaphragmatic dysfunction is a major reason for difficulty in liberation from MV and may be present in up to 80% of patients with difficult or prolonged weaning (12). Difficult weaning is defined as successful weaning between the second and sixth days after the first attempt; prolonged weaning is defined as success after more than six days or persistent ventilator dependence after multiple attempts. Both are associated with worse prognosis and higher mortality (16.5% to 29.8%) compared to simple weaning, which occurs within 24 hours of the initial attempt (13). Diaphragmatic dysfunction may also result in long-term ventilator dependence, requiring home ventilation or transfer to care facilities (14).

Strategies to prevent or reverse diaphragmatic weakness during MV increasingly focus on inspiratory muscle training. The most common approach is to apply an external load to the respiratory muscles during inspiration to improve the strength and endurance of both the diaphragm and the accessory inspiratory muscles (15). Randomized trials and systematic reviews have demonstrated that inspiratory muscle training can increase inspiratory muscle strength in mechanically ventilated patients as measured by maximal inspiratory pressure and diaphragm ultrasonography. However, evidence for its effects on outcomes such as MV duration, time to liberation from MV, intensive care unit (ICU) stay, and mortality remains inconsistent (16,17). Therefore, in practice, inspiratory muscle training may be considered an adjunct to structured weaning protocols, especially for patients needing prolonged MV or with difficult weaning.

An alternative strategy to prevent or attenuate diaphragmatic atrophy during MV is diaphragmatic neurostimulation. This technique involves placing electrodes on or near the phrenic nerve, or directly on the diaphragm at the phrenic nerve’s insertion site. Neurostimulation enables diaphragmatic contraction during MV, including under controlled modes of ventilation (18). The development of noninvasive and minimally invasive neurostimulation modalities—obviating the need for surgical electrode implantation—has expanded the feasibility of this therapeutic approach in critically ill patients receiving MV. Noninvasive options include cervical magnetic or electrical stimulation, whereas minimally invasive approaches comprise transcutaneous electrical stimulation (cervical or diaphragmatic) and transvenous diaphragmatic neurostimulation (18,19). Among these modalities, transvenous diaphragmatic neurostimulation has been the most extensively investigated in clinical studies involving mechanically ventilated patients.

Theoretically, the primary rationale for transvenous diaphragmatic neurostimulation during MV is to mitigate muscle atrophy resulting from diaphragmatic inactivity. In this context, the principal indication for neurostimulation would be in patients undergoing prolonged MV and exhibiting muscle weakness, with the aim of improving diaphragmatic strength. A secondary potential indication is its earlier application to prevent diaphragmatic atrophy (20,21).

Additional potential benefits of transvenous diaphragmatic neurostimulation have been proposed. In patients with severe acute respiratory distress syndrome (ARDS) receiving controlled MV, the introduction of cycles incorporating diaphragmatic contraction may promote a more homogeneous distribution of tidal volume, particularly toward the dependent (dorsal) lung regions. This may allow ventilation at lower alveolar pressures, an important protective factor against ventilator-induced lung injury, as well as improve ventilation–perfusion matching, thereby enhancing oxygenation (22). Concurrently, diaphragmatic contraction during MV may increase venous return and mitigate the hemodynamic effects associated with positive-pressure ventilation (23).

The clinical impact of these physiologic benefits of diaphragmatic contraction induced by neurostimulation remains to be established. The first randomized clinical trials have been conducted using transvenous diaphragmatic neurostimulation. Dres and colleagues conducted the RESCUE-3 trial, which aimed to evaluate the efficacy and safety of transvenous diaphragmatic neurostimulation for weaning patients from invasive MV (24).

This was a multicenter, open-label randomized clinical trial conducted across 48 centers in the United States and Europe. Eligible participants were adults (>18 years of age) who had been receiving MV for more than 96 hours and who, despite meeting predefined readiness-to-wean criteria, had experienced more than two failed attempts at liberation from MV. At least one of these attempts required a spontaneous breathing trial performed with pressure support of 0 to 5 cm H₂O and PEEP of 0 cm H₂O.

Patients were randomized in a 1:1 ratio to receive diaphragmatic neurostimulation via a catheter inserted into the left subclavian or left internal jugular vein or to a control group. Diaphragmatic neurostimulation sessions were conducted twice daily (6 sets per session, with 10 stimulations per set), with a minimum interval of 3 hours between sessions. During each session, stimulation pulses were delivered at 15 Hz, with a pulse duration of 200–300 µs and an intensity of ≤27 mA. Both groups received a standardized protocol for sedation minimization and early mobilization. In addition, patients were assessed daily for readiness-to-wean criteria, and, if these were met, a standardized weaning attempt was performed on that day.

The primary outcome was weaning success at day 30, defined as remaining free from MV for more than 48 hours following disconnection. For the analysis of this outcome, the investigators used a Bayesian adaptive model and incorporated data from the RESCUE-2 (25) trial to inform the prior distribution. Interim analyses were planned after every additional 50 patients enrolled, up to a maximum of 400 participants. At each interim analysis, the posterior probability of treatment superiority for successful weaning (hazard ratio >1) was calculated. If this probability exceeded the prespecified thresholds (98.25% to 99.25%), superiority would be declared, and the trial would be stopped early. Despite this design, the trial was discontinued after enrolling 223 patients (109 in the intervention group and 114 in the control group), as slow recruitment failed to meet the prespecified posterior probability threshold for superiority.

By day 30, 71 patients (70%) in the diaphragmatic neurostimulation group and 69 patients (61%) in the control group had successfully weaned. For the primary outcome, the posterior probability of superiority of the intervention for weaning success was 97.9% [adjusted hazard ratio, 1.34; 95% confidence interval (CI), 1.01 to 1.78]. These findings were consistent across sensitivity analyses, including when successful weaning was defined as remaining free from MV for more than 7 days and when the intention-to-treat population was considered. With respect to secondary outcomes, the duration of MV until successful weaning or day 30 was shorter in the intervention group (difference, −2.5 days; 95% CI, −5.0 to 0.1), whereas 30-day mortality was similar between groups (9.8% vs. 10.5%; posterior probability of superiority, 80.6%). Maximal inspiratory pressure improved to a greater extent in the neurostimulation group (adjusted difference, 5.9 cm H₂O; 95% CI, 1.8 to 10.0 cm H₂O; posterior probability of superiority, 99.8%).

These results suggest that diaphragmatic neurostimulation may enhance diaphragmatic contractile strength and increase the likelihood of successful weaning from MV in patients with difficult weaning. However, several important limitations should be considered when interpreting these findings. As an open-label trial, decisions regarding ventilator discontinuation may have been influenced by knowledge of group allocation. In addition, due to slow recruitment, the final sample size was smaller than planned, precluding attainment of the prespecified posterior probability threshold for superiority.

When the results of this trial are compared with those of the RESCUE-2 study (used to inform the prior), which was conducted by the same investigators with similar objectives, important methodological differences emerge. Regarding inclusion and exclusion criteria, in RESCUE-3, the investigators excluded patients with a maximal inspiratory pressure greater than 50 cm H₂O, thereby selecting a population more likely to derive benefit from diaphragmatic neurostimulation. Regarding outcomes, RESCUE-2 did not demonstrate superiority of diaphragmatic neurostimulation over the control group. The cumulative incidence of successful weaning from MV by day 30 was 82% in the intervention group and 74% in the control group, corresponding to an absolute difference of 7% (95% CI, −10% to 25%; P=0.59).

In absolute terms, these estimates were not substantially different from those observed in RESCUE-3. The principal distinction between the two trials lies in the statistical framework: RESCUE-3 employed a Bayesian approach, whereas RESCUE-2 used a conventional frequentist analysis. Had the RESCUE-3 data been analyzed within a frequentist framework, the findings would likely have mirrored those of RESCUE-2—namely, no demonstrable benefit of diaphragmatic neurostimulation (71 patients achieved successful weaning in the intervention group vs. 69 in the control group). Moreover, the declaration of superiority in RESCUE-3 was based exclusively on the direction of the effect (hazard ratio >1), as no prespecified threshold for the magnitude of effect was defined. This limits the interpretation of the clinical relevance and practical significance of the observed findings.

Another important consideration regarding transvenous diaphragmatic neurostimulation is its safety profile, despite being classified as a minimally invasive procedure. In RESCUE-3, the rate of serious adverse events was higher in the intervention group (35.8%) than in the control group (23.7%). Among patients who experienced serious events potentially related to the procedure, two deaths were reported—one due to hemothorax/pneumothorax and another due to an acute coronary syndrome precipitated by tension pneumothorax.

Weaning from MV remains a major challenge in the management of critically ill patients, particularly among those requiring prolonged ventilatory support and who have already developed muscle weakness. In this context, strategies targeting key determinants of weaning failure, such as diaphragmatic atrophy and weakness, are of considerable interest. However, successful weaning is a multifactorial process influenced by patient-related factors (e.g., age, comorbid conditions, and the underlying cause of respiratory failure), complications acquired during the ICU stay (including infections, malnutrition, muscle weakness, and fluid overload), and organizational aspects of care (such as the implementation of sedation and weaning protocols, team expertise, and ICU infrastructure). Consequently, the impact of isolated interventions is likely to be limited, underscoring the need to integrate them into comprehensive, protocolized care strategies. Transvenous diaphragmatic neurostimulation may represent one such adjunctive intervention. Nonetheless, its role and safety in the complex management of prolonged weaning require further evaluation in adequately powered, methodologically rigorous studies.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.