Assisted mechanical ventilation promotes recovery of diaphragmatic thickness in critically ill patients: a prospective observational study

Diaphragm dysfunction during mechanical ventilation is increasingly recognized as a clinical entity which has an impact on important clinical outcomes, such as the risk of prolonged mechanical ventilation, re-intubation, tracheostomy, and death [1]. Ultrasound (US) is becoming a reference to assess diaphragmatic injury and to detect changes in diaphragmatic thickness over time; both an increase and a decrease of diaphragmatic thickness are associated with diaphragm activity and ventilation outcome [1, 2]. Starting from experimental and clinical studies, four mechanisms of diaphragmatic injury (myotrauma) have been identified: disuse atrophy due to over-assistance [3], excessive load due to under-assistance [1, 4], eccentric myotrauma from diaphragm contraction during expiration [5] or asynchronies, and longitudinal atrophy, due to high level of positive end-expiratory pressure (PEEP) [6]. Therefore, the need to set “protective” mechanical ventilation not only for the lungs but also for the diaphragm was recently advocated [7]. The targets of this type of ventilation would be to avoid excessive or minimal diaphragm activity, trying to mimic the inspiratory effort of healthy subjects [8]. In fact, a diaphragmatic thickening fraction between 25 and 40% (considered the reference range for resting tidal breathing) was associated with stable diaphragmatic thickness [2], shorter duration of mechanical ventilation [1], and higher incidence of weaning success [9]. Given all of these, it is advisable to strictly monitor patients’ inspiratory efforts, and specifically diaphragmatic function, while avoiding lung injury [10].

Some clinical studies showed that intermittent diaphragm stimulation during cardiothoracic surgery could exert benefits on diaphragmatic force generation and mitochondrial function [11, 12]. On the other side, experimental studies showed that the mode of ventilation could influence the degree of diaphragmatic dysfunction. In healthy rats, controlled mechanical ventilation (CMV) lead to proteolysis and decreased protein synthesis, while the same period spent under pressure support ventilation (PSV) preserved diaphragmatic protein content, with no increase in oxidative injury [13]. In another animal study, assisted mechanical ventilation (AMV) prevented atrophy and loss of diaphragmatic force caused by controlled ventilation [14]. Recently, the first study conducted on humans (two groups of organ donors as study groups and thoracic surgery patients as controls) was published showing a correlation between the severity of diaphragm histological damage and the mode of ventilation used. In particular, the duration of the period without diaphragm stimuli (i.e. CMV) correlated with a decreased cross-sectional area of the fibres in both study groups compared to controls [15]. One of the two study groups underwent a longer period of CMV than the other organ donors group, but this did not lead to a significant reduction in diaphragm fibre cross-sectional area. The same group spent more time also under an assisted mode of ventilation, with some preservation of diaphragmatic function, and this was advocated as the possible mechanism which avoided excessive atrophy.

The aim of this study is to test the hypothesis that, in the face of decreased diaphragmatic thickness during CMV, assisted modes of ventilation which partially preserves diaphragmatic activity could restore diaphragmatic thickness. Secondarily, we aim to correlate the change in diaphragm thickness during both CMV and AMV to clinical factors and outcomes. As the third objective, we aim to describe the boundaries of “diaphragm-protective assisted ventilation” according to the clinical outcomes of our patients and to correlate diaphragm activity to thickness.

This is a prospective observational study performed in the general ICUs of San Gerardo Hospital in Monza and Desio Hospital, Cernusco S/N Hospital, Niguarda Hospital, and Ospedale Maggiore Policlinico in Milan.

The study protocol was approved by the local Ethics Committees of the involved hospitals, and informed consent was obtained from patients or their substitute decision-makers prior to enrolment.

The participating centres managed patients in the acute phase of the critical illness with CMV (with or without muscle paralysis but avoiding spontaneous breathing effort), switching to AMV after improvement of the clinical condition (i.e. resolution of acute phase, FiO₂ < 50%, reduction in the doses of vasopressors), typically by PSV. PS level was progressively reduced until a phase of continuous positive airway pressure (cPAP) ventilation. In this context, it is possible to clearly separate the fully controlled or assisted ventilation mode and to identify the onset of inspiratory activity.

While diaphragm atrophy is widely described in the recent literature, the reversal of this process is still poorly studied and it is not clear which factors can favour it. Moreover, the increase of diaphragm thickness described during some phases of mechanical ventilation might not represent a recovery of the muscle mass but the occurrence of a pathological process. The purpose of this study was to partially address these open questions by looking at the effect of assisted mechanical ventilation on diaphragm thickness and activity and their relationship to patients’ outcome.

The main results can be summarized as follows: diaphragm thickness is influenced by the mode of ventilation; in fact, while during CMV, Tdi tends either to decrease or not change, spontaneous assisted breathing with pressure support can lead to partial restoration of Tdi. A decrease in Tdi during CMV is associated with a longer period of diaphragm inactivity (duration of CMV), while a regain in Tdi during AMV in our population was not associated with common clinical outcomes. Considering diaphragm activity, the patients with lower TFdi trended towards less AMV days. Lastly, we found a correlation between the increase in Tdi during AMV and lower respiratory rate and lower Rapid Shallow Breathing Index as well as higher Pressure Muscle Index, whereas we found no association between diaphragm thickness and activity, as measured by TFdi.

Our data confirm that diaphragm atrophy occurs in ICU patients, and this can be captured by US [27]. The mean thickness value recorded at the study inclusion was similar to that recorded in another study [25]. A decrease in Tdi was shown to be associated with longer periods of mechanical ventilation, and it occurs early after intubation [2, 25]. The higher levels of PEEP in patients with diaphragm thinning during CMV could be in line with the recent evidences regarding myofibre longitudinal atrophy due to PEEP [6]. To our knowledge, no studies until now described the relation between the prealbumin levels, as an index of nutritional status [17], and diaphragm thickness. If it can be argued that both lower prealbumin and diaphragm thinning are simply a marker of more severe baseline disease, this is not confirmed by differences in baseline SOFA or incidence of shock in the two groups (thinning or unchanged diaphragm during CMV). Therefore, these observational data can give a suggestion regarding a possible preventive approach to avoid diaphragm thinning by means of monitoring and optimization of nutritional status.

While the negative effects of prolonged diaphragm inactivity are widely recognized, clinical data regarding the consequences of partial diaphragm activity, which happens during assisted modes of ventilation, are less clear and somehow conflicting.

Animal studies clearly showed that different modes of assisted ventilation induce a milder loss of diaphragm function [14, 28] while also causing less proteolysis [13].

On the clinical side, instead, the first important factor to be underlined is that, because of the ventilation practice of different centres, the period of controlled and assisted ventilation are not clearly dissectable, and therefore, it is difficult to correlate the trend of diaphragm thickness with a specific mode of ventilation [25]. Two Italian studies focused specifically on the consequences of the mode of ventilation on the diaphragm. Umbrello et al. [29] showed that the level of diaphragm activity was related to the level of support provided by the ventilator and to other indices of respiratory effort (p0.1, PMI), but given the short duration of patient observation in this study, it lacks a trend of Tdi measurements overtime. Zambon et al. [26] instead correlated Tdi with the ventilation setting and, in keeping with our results, demonstrated that the daily atrophy rate of the diaphragm is directly related to the mode of ventilation; partial preservation of diaphragm activity during PSV or spontaneous breathing with cPAP significantly reduced the atrophy rate. In 2015, Goligher et al. [2] showed a direct relation between diaphragm thickness and activity, expressed as TFdi. In a supplementary analysis of this study, patients were classified based on the mode of ventilation (controlled versus partially assisted): in both cases, the diaphragm thickness decreased at the beginning of ventilation time, then it tended to increase only in the “partially assisted mode” patients, despite not significantly. The relation between diaphragm Tdi, diaphragm activity, and duration of mechanical ventilation shown in Goligher’s study [2] was not confirmed by our results; this could be due by the fact that, on average, the Tdi in our patients tended to decrease and then to return to baseline. An intriguing very recent study [15] shows that diaphragm partial activity during mechanical ventilation preserves muscle fibres from histological damage, contrarily to muscle complete inactivity. We have no histological data on patients’ diaphragms, and further studies, maybe involving muscular structure evaluation, are needed to judge if the muscle actually benefits from the assisted activity, even if in some cases, this might not translate in improved outcome due to the co-existence of too many confounding factors in an ICU patient.

Regarding diaphragm activity, some studies report that the level of TFdi can be used as a predictor of successful weaning [9, 30, 31]. This evidence was not confirmed in other studies. One of them found TFdi to be less sensitive and specific of traditional measure of breathing effort and distress [32]. Another recent study [33] looked specifically at extubation failure in those patients who successfully passed a spontaneous breathing trial (SBT). In this population, neither TFdi nor diaphragm excursion was predictors of successful extubation, and the only factor associated with successful extubation was effective cough. Sixty-three percent of the patients who passed the SBT and were successfully extubated had a TFdi of the right diaphragm lower than 30%.

Similarly, we found low values of mean and maximum TFdi during the period of AMV in our patients. First, we focused on tidal TFdi, while in other studies, TFdi is calculated between maximum inspiration and expiration [2, 34, 35]. Second, in a larger population of ventilated patients, the majority of the subjects (128 out of 146) had a TFdi lower than 25% during the first 3 days of ventilation [1]. Third, the low TFdi that we recorded somehow reflects the weaning approach used in the centres involved in the study. All the centres are geographically close and traditionally use PSV as the main mode to wean patients from a mechanical ventilator. Spontaneous breathing trials are usually performed in a cPAP mode with a level of PEEP < 10 cmH₂O and only after a gradual decrease of the level of pressure support, which could happen along many days. The amount of inspiratory effort is usually measured more than once per day, by means of p0.1 and PMI. PEEP is typically set by the clinician based both on respiratory system compliance and oxygenation response [36], taking also into account that the level of PEEP itself can modify the respiratory effort of the patient [37]. Therefore, it is very uncommon that patients’ endurance is tested causing strenuous levels of inspiratory effort, as also witnessed by the low median Rapid Shallow Breathing Index in all the patients. TFdi is normally not monitored by the clinician. These results could help delineate the parameters and safe thresholds for “diaphragm-protective mechanical ventilation”, and once again, they underline the importance of monitoring the level of inspiratory effort in spontaneously breathing patients and the need to avoid strenuous efforts. In fact, a trend towards longer mechanical ventilation and longer ICU stay was observed in those patients whose mean TFdi was higher in the first days of the weaning process. This could suggest that keeping the diaphragm at rest, with low levels of TFdi, promotes rehabilitation of the muscle. Or that, in patients who are healthier, and therefore ventilated for a shorter time, a low TFdi is enough to promote successful weaning from the mechanical ventilator.

RSBI is a commonly used index to evaluate extubation readiness during weaning [20], whereas PMI (namely the pressure recorded on the ventilator during an inspiratory hold in PSV) was shown to tightly correlate with inspiratory pressure generated by the patient’s muscles during inspiration [19]. We found a correlation between an increase in diaphragm Tdi during AMV and lower RSBI; the presented data do not allow to infer any causal relationship between these two phenomena, but for the first time, we show that a respiratory pattern as close as possible to normal breath happens at the same time as a partial restoration in diaphragmatic thickness. On the other side, the correlation between recovery of thickness and PMI could support the hypothesis that muscle thickness restoration brings to a higher force generation, leading to a trend towards higher tidal volumes despite lower mean pressure support levels (Table 3). All these data point towards the fact that diaphragm thickness recovery during AMV is a different process when compared to diaphragm absolute increase in thickness, as described by Goligher et al. [2], the latter correlated with diaphragm dysfunction. In our patient, restoration of diaphragm thickness through a smooth training during AMV was associated with a more efficient diaphragm function. This observation could be hypothesis-generating for a subsequent study.

The main strength of this study lies in the ability of clearly separating the phases in which patient’s inspiratory effort is present or absent, enabling to investigate the specific consequences of the mode of ventilation on diaphragm thickness and activity. We acknowledge that the consequences of ventilation on myotrauma need to be further analysed [38], even if a “lung-protective ventilation” during assisted modes, which mainly consists in the strict monitoring of respiratory mechanics and effort is the main key to improve outcome [10] bringing also to a smooth training of the diaphragm while weaning the lungs.

The mode of ventilation affects diaphragm thickness in ICU patients. While diaphragm atrophy is associated with a longer duration of controlled mechanical ventilation, assisted mode leads to partial restoration of diaphragm thickness, which seems to correlate to a better breathing pattern. These findings on one side underline the need for monitoring respiratory mechanics during spontaneous breathing and avoiding excessive efforts, on the other side provide some data to define the thresholds for “diaphragm-protective mechanical ventilation”.