FormalPara Take-home message

In elderly patients with high risk of weaning or extubation failure, lung ultrasound at end of SBT can aid in predicting weaning outcome.

Introduction

The weaning process represents 40–50% of total duration of mechanical ventilation. Whatever its primary cause, the switch from mechanical ventilation (MV) to a spontaneous breath trial (SBT) induces significant changes in lung aeration [1]. Lung ultrasound (LUS) is a reliable tool to detect lung aeration and can be used to reliably predict extubation failure [2]. Since failure in weaning process is associated with poor outcome [3], a reliable prediction of extubation failure and the identification of its underlying mechanism may be of importance on prognosis. For example, diastolic dysfunction and the subsequent weaning-induced pulmonary oedema (WIPO) could be effectively treated by fluid removal or vasodilators [4]. Pulmonary oedema results in increased extra-vascular lung water (EVLW) that extends to lung periphery and is detected by LUS as vertical ultrasound artefacts called B-lines. The B-line number is correlated to pulmonary wedge pressure [5]. Variation of this number is correlated to variation of EVLW during haemodialysis [6] or stress test [7]. Thus B-line assessment to evaluate pulmonary oedema during heart failure is now recommended [8].

During a spontaneous breathing trial (SBT), a prevailing A-line pattern changing to a diffuse B-line pattern is suggestive of cardiogenic oedema as the cause of weaning failure [9]. Recently, a number of B-lines ≥ 6 on four anterior points during SBT has been proposed as diagnostic test for WIPO [10]. Although the diffuse bilateral B-lines indicate interstitial syndrome and increased lung density [1], it is not specific for cardiogenic pulmonary oedema, given that loss of aeration or de-recruitment can also create this LUS finding.

Transthoracic echocardiography (TTE) can be used to estimate left ventricular filling pressure [11]: if performed before and during a spontaneous breathing trial (SBT), an increase in pre-load pressure is identified by increase in E/A and in E/E′ [12].

We hypothesized that the addition of echocardiography to LUS in selected elderly critically ill patients with isolated diastolic left ventricle (LV) dysfunction could help identify the risk of weaning failure as well as understand the cause. So, we conducted a prospective observational study to estimate the value of LUS alone, TTE alone, and LUS and TTE in combination to 1/ to identify elderly patients at high risk of weaning failure from MV, and 2/ to determine if LUS alone could identify patients at risk of weaning failure.

Methods

This prospective, observational study was conducted in four teaching hospital ICUs (CHU Dijon, France, St. Joseph Hospital, Paris, France, Fondazione IRCCS Policlinico S. Matteo, Pavia, Italy and King Abdulaziz Medical City, National Guard Health Affairs, Riyadh, Saudi Arabia). This study was approved by local Institutional Ethics Committees and registered with ClinicalTrials.gov No. NCT03261440. Ultrasound lung and cardiac examination were a part of a routine diagnostic strategy aimed at optimizing respiratory and haemodynamic status of the patients. So, this observational study was classified as a study of standard care and, therefore, the need for informed consent was waived. Oral information was provided to the patient or relatives. No patient or family member expressed concern about participation.

Inclusion criteria: patients aged > 65 years, LV ejection fraction > 50% at enrolment on the apical four-chamber view, mechanically ventilated for more than 48 h and eligible for their first SBT according to the attending physician’s judgement that the underlying disease which led to intubation had sufficiently resolved.

Exclusion criteria: tracheostomy, known neuromuscular pathologies, ejection fraction < 50% at enrolment, significant COPD history, previous SBT failure.

Ultrasound examination

All ultrasound examinations were conducted using a CX-50 Philips or Vivid-iTM GE Healthcare and included both a lung and cardiac examination.


Lung ultrasound: A comprehensive scan was taken in six regions for each lung (superior and inferior areas in the anterior, lateral, and posterior fields using anterior and posterior axillary lines as landmarks). Four ultrasound aeration patterns were defined in each region [13]: (1) normal aeration (N—score 0): presence of lung sliding with A lines or fewer than two isolated B lines; (2) moderate loss of lung aeration: multiple, well-defined B lines (B1 lines—score 1); (3) severe loss of lung aeration: multiple coalescent B lines (B2 lines—score 3); and (4) lung consolidation (C—score 3), presence of a tissue-like pattern. For a given region of interest, points were allocated according to the worst ultrasound pattern observed. Global, anterior and antero-lateral LUS scores were calculated as the sum of all, anterior and antero-lateral regions ranging, therefore, 0–36, 0–12 and 0–24, respectively.


Cardiac ultrasound: All the examinations were made on the apical 4-chamber view. Presence of a left ventricular (LV) systolic dysfunction was assessed by eye-ball evaluation (ejection fraction ≤ 50% with LV dilatation). The transmitral flow was recorded by pulsed Doppler with the sample volume placed at the mitral valve tips. The mitral inflow was analysed for peak velocity of early (E) and late (A) filling. Velocities of the mitral annulus were recorded using tissue Doppler imaging (TDI) program with a 5-mm sample volume placed at the lateral corner of the mitral annulus [11]. The early diastolic (Ea) velocity of mitral annular displacement was measured from the TDI recording.

Definitions

Filling pressures: LV filling pressure can be evaluated using Doppler study of mitral flow and Doppler tissue imaging [11].

Low filling pressures or impaired relaxation patterns were defined by the presence of E/A ≤ 1 and E/Ea ≤ 13 [11]. High filling pressures or pseudonormal patterns were defined by E/A > 1 and E/Ea > 13 [11]. High filling pressures with restrictive patterns were defined by a E/A > 2.

Weaning-induced pulmonary oedema was defined by. an antero-lateral LUS score ≥ 5 with an increase in the LV filling pressure during a SBT. Derecruitment was defined by an increase in LUS scores without an increase in LV filling pressures during the SBT.


Weaning outcome: Weaning failure was defined by the failure of SBT [14]. Extubation failure was defined by a need for reintubation, non-invasive ventilation or death within 48 h after extubation.

Protocol

The SBT was performed through a T-tube as previously described or under mechanical ventilation with PEEP ≤ 5 cmH2O and pressure support of ≤ 7 cmH2O. The type of SBT performed was chosen by the physician in charge of the patient independently of the protocol. LUS examination and mitral Doppler study were performed prior to SBT under mechanical ventilation and at the end of SBT. The decision to stop SBT or to extubate was made by the physician in charge independently of the investigators.

Statistical analysis

In this interim analysis study, the estimated sample size to distinguish failing and non-failing patients on the basis of lung ultrasound score [2] [17 ± 3 vs. 13 ± 3], expecting a 30% failure rate and a 10% dropout, with alpha error 0.05 and power 0.95, was 38. So, we included 40 patients. Results are expressed as median (interquartile range) for quantitative variables and number and percentage for categorical ones. Two comparisons were performed: (1) patients successfully extubated (weaning success) vs. patients with weaning failure or extubation failure and (2) patients with weaning success vs. patients with extubation failure. Unpaired Wilcoxon/Mann–Whitney U-test for numerical data and the Fisher’s exact test for categorical ones were used. ROC curves analysis was performed to study ability of LUS score, anterolateral LUS score, E/A at end of SBT to predict weaning failure, and/or extubation failure. Sensitivity specificity, positive/negative predictive values, and positive/negative likelihood were also calculated. Cut off points for multivariate analysis were obtained by Youden index. To evaluate the association between endpoint variable (development of extubation/weaning failure) and ultrasound features, odds ratios (ORs) and the corresponding 95% confidence interval (CI) were calculated; a logistic regression model was carried out to identify the mutually adjusted effect among endpoints and the independent variables chosen on the basis of the statistical significance (univariate analysis, p ≤ 0.05) and of the clinical judgment; age and sex were used as adjusting variables. P-value ≤ 0.05 was considered significant (two-sided). All the analyses were conducted with STATA/SE for Macintosh, version 14.2.

Results

Patients

Flowchart of the study is given in Fig. 1. 40 patients were included (consecutively in each unit) in the study: eight patients were determined to have weaning failure and therefore remained intubated and 32 patients were extubated. Among them, ten patients presented an extubation failure and the remaining 22 were successfully weaned. Clinical characteristics are given in Tables 1 and 2. Patients’ median age was 72 [67–78] years, mechanically ventilated for a median duration of 8.5 [6.0–15.5] days before the first weaning attempt, and with significant cardiac comorbidity (20.0% of patients were admitted after cardiac surgery or for cardiogenic pulmonary oedema). Consequently, a high incidence of 45% (IC 95%: 28–61) of both weaning + extubation failure was observed. No significant differences were found between groups for clinical characteristics.

Fig. 1
figure 1

Weaning Outcomes

Full size image
Table 1 Clinical characteristics for all patients
Full size table
Table 2 Clinical characteristics of extubated patients
Full size table

LUS for predicting weaning or extubation failure

No significant difference was found between groups in LUS score prior to SBT (Tables 3 and 4). Significant differences in all LUS scores were found between failure groups vs. success group at the end of SBT (Tables 3 and 4). ROC analysis for ability of SBT LUS scores to predict weaning/extubation failure and extubation failure are given in Fig. 2a, b, respectively.

Table 3 Ultrasound characteristics prior to and at end of a SBT for all patients
Full size table
Table 4 Echocardiography characteristics prior to and at end of a SBT for extubated patients
Full size table
Fig. 2
figure 2

Ability of lung ultrasound (LUS) to predict weaning or extubation failure (a) or extubation failure (b) at end of spontaneous trial (SBT). Receiver operating characteristic curves of global, anterior and antero lateral LUS scores are shown

Full size image

Cardiovascular abnormalities before and during spontaneous breathing trial

No significant difference was found between groups in echocardiographic parameters either during or prior to SBT. ROC analysis for E/A and E/Ea to predict weaning/extubation failure and extubation failure are given in Fig. 3a, b, respectively. Mitral flow pattern during MV and at end of SBT (i.e. degree of diastolic dysfunction: impaired relaxation, pseudonormal and restrictive patterns) were not significantly different between groups (Tables 3 and 4). A mitral pattern variation ≥ 1 step was observed overall in 15.0% of patients but was not associated with failure (Tables 3 and 4).

Fig. 3
figure 3

Comparison of ability of lung ultrasound scores (LUS) and E/A and E/Ea ratios to predict weaning or extubation failure (a) or extubation failure (b) at the end of SBT. Receiver operating characteristic curves of E/A, E/Ea and globall lung ultrasound scores are shown

Full size image

LUS and mitral flow during weaning process and WIPO

At the end of the SBT, an antero-lateral LUS score ≥ 5 was found in 30 patients and in 21 this was associated with high filling pressure. An anterolateral LUS score ≥ 5 was associated with high prevalence of weaning or extubation failure (56.7%), independently from LV filling pressure: in patients with anterolateral LUS scores ≥ 5 and high filling pressure, extubation or weaning failure had a similar prevalence compared to patients with antero-lateral LUS ≥5 and low LV filling pressure (57.1% vs. 55.6%; p = 1.0000). Among patients with antero-lateral LUS scores ≥ 5 and high LV filling pressure, only three could be classified as WIPO, i.e. an increase of LV pressure was induced by the switch from MV to SBT (Fig. 4).

Fig. 4
figure 4

Left ventricular filling patterns and Lung ultrasound anterolateral scores at end of spontaneous trial (SBT) and weaning outcome for all included patients. MV mechanical ventilation, WIPO Weaning induced pulmonary oedema

Full size image

Moreover, there was no correlation between LUS scores and diastolic pattern: some patients with increased LV pre-load, as measured by E/Ea > 13, had low LUS scores, while other patients with normal LV-preload had high LUS scores (Fig. 5a, b).

Fig. 5
figure 5

Correlation between Global (a) and anterolateral (b) Lung ultrasound (LUS) score and E/Ea ratio at end of spontaneous trial (SBT). No significant correlation was found

Full size image

In multivariate regression analysis, only global and anterolateral LUS scores at the end of the SBT were independent risk factors for weaning and / or extubation failure (Table 5).

Table 5 Multvariate analysis for risk factors associated with difficult weaning outcome
Full size table

Discussion

This study found that in elderly patients, only increased LUS (global or anterolateral) score was an independent risk factor for weaning or extubation failures, while increased LV filling pressure measured by echocardiography during SBT was not associated with weaning or extubation failures. Our results show that increases in LV filling pressure may occur without inducing pulmonary oedema. Conversely, during the weaning process, antero-lateral LUS score may occur without an increase in LV filling pressure. LUS identifies a lung loss of aeration induced by both WIPO and other mechanisms, such as derecruitment; therefore, an increase in LUS scores was not specific for WIPO but instead accurately identified the patients at risk of weaning/extubation failure.

SBT is associated with changes in aeration. If significant, these changes in lung aeration can be detected using LUS [15, 16]. The best way to semi-quantify loss of aeration at the bedside is to use a LUS rating system [2, 13, 17, 18]. In the context of weaning, the LUS score can help distinguish patients at high and low risk of post-extubation distress. In patients passing a 1-hour SBT, the LUS score is predictive of extubation success or failure: in patients successfully extubated, overall lung aeration does not significantly change during the SBT, whereas in patients with extubation failure, lung aeration decreases during the SBT. After a successful SBT, a score > 17 was highly predictive of extubation failure, while a score < 13 was highly predictive of weaning success in an unselected ICU population [2]. We obtained very similar results in elderly patients at the first attempt to wean from mechanical ventilation.

LUS is simple and easy to learn [19]. The change from dominant A-lines to diffuse B-lines during an SBT is easy to diagnose with LUS for the majority of ICU physicians and corresponds to an increase in lung tissue density in the explored region. LUS score predicts extubation failure better than other ultrasound techniques presumably because LUS detects any increase in tissue density, independently from its aetiology. The decrease in pulmonary aeration measured by LUS score is the final common pathway of different mechanisms inducing weaning or extubation failure, namely cardiogenic pulmonary oedema and derecruitment. Hence, LUS could be a key tool in the ICU during weaning process.

Switching a patient from a controlled ventilation mode to a SBT is a stress test for the ICU patient and can induce cardiogenic pulmonary oedema in cases of volume overload as well as LV systolic or diastolic dysfunction. This can be diagnosed through advanced cardiac echography skills and Doppler study of mitral flow and Doppler tissue of mitral annulus [11, 20]. Moreover, cardiac echography may also be used also to diagnose which cardiogenic cause is responsible for this failure [9]. As expected, in our study including elderly patients, when the pre-test probability of cardiac weaning failure is high, cardiac echography has limited utility [20]. When the pre-test probability is low or intermediate, as in an unselected population of critically ill patients, cardiac ultrasound might be more useful in predicting weaning failure [21,22,23].

LUS may help in detecting WIPO. LUS pattern of cardiogenic pulmonary oedema is characterized by the presence of multiple B-lines in all examined regions [24] since alveolar flooding is homogeneously distributed. Simple sum of B-lines yields a score correlated to the extent of extravascular lung water (EVLW) and was correlated to surrogate markers of pulmonary oedema as EVLW measured by transpulmonary thermodilution or pulmonary wedge pressure [5, 7]. Variation of EVLW during hemodialysis [6] or stress test [7] also correlates with changes in the number of B lines, and their use in the assessment of pulmonary oedema during heart failure is now recommended [8]. Recently, a number of B-lines ≥ 6 on 4 anterior points during SBT has been proposed as a diagnostic test for WIPO [10].

But, as above, B-lines alone are not enough to affirm WIPO. In fact, loss of lung tissue aeration associated with B-lines is not specific of cardiogenic pulmonary oedema, but it can be alternatively explained by lung derecruitment during SBT. Moreover, our findings suggest that the combination of B-lines and high filling pressure are also not specific for WIPO. During an increase in left atrial pressure or in hypervolemia, there is an increase in pulmonary venous pressure and a more homogeneous redistribution of perfusion to the apices by distension and recruitment of capillaries, and an increase of the lymphatic flow in parallel. All these mechanisms allow limiting the increase in capillary pressure and oedema formation [25, 26]. This was observed in heart failure patients where haemodynamic congestion (i.e. an increase of filling pressure) is not always associated with clinical congestion [8] even in cases of severe haemodynamic congestion [27, 28]. We observed similar findings in a population of elderly patients undergoing their first SBT: high filling pressures were not synonymous with cardiogenic pulmonary edema in this setting.

The combination of lung ultrasound with other ultrasound techniques in the assessment of the patients to be weaned has been suggested, not only to identify the frailest patients early, but also to understand the underlying mechanism, since the main causes of weaning failure can be evidenced by ultrasound (i.e. unresolved lung disease, diaphragm dysfunction and cardiac failure) [9, 29]. A combined cardiac, lung and diaphragm ultrasound approach may offer a more comprehensive understanding of patient’s status [20] and can potentially fill the gap of single-organ assessment.

Our findings suggest a 2-step ultrasound-aided strategy to improve the weaning process: (1) in all patients, LUS assessment at the end of SBT is required to identify those patients at high risk of extubation failure, and (2) in the weaning failure group, cardiac and diaphragm ultrasound could be used to identify the mechanism and guide treatments to promote extubation success.

Our study has some limitations. First, this was a multi-centre study, with only 40 patients included. Second, the diagnosis of weaning-induced pulmonary oedema was not based on pulmonary artery catheter or on a review by experts. However, echocardiogragraphy and Doppler studies are validated tools to diagnose increased LV pre-load. Third, we tested mainly patients with SBT performed with T-tubes, but for ten patients low pressure support was used which is not as challenging for the heart [30] as spontaneous breathing with T-tube.

In conclusion, our study shows that LUS, but not echocardiography, can predict extubation outcome. New studies are needed to test the hypothesis that LUS, by allowing for a more comprehensive assessment of the patients at risk for extubation failure, can also improve the weaning outcome.