Myocardial ischemia during ventilator weaning: a prospective multicenter cohort study

This prospective multicenter cohort study was performed in four medical or mixed intensive care units of secondary and tertiary hospitals in France from February 2012 to May 2016. Patients screened for enrolment were those intubated for at least 24 h with ventilator settings allowing to initiate the weaning process [SpO₂ > 90% or PaO₂/FiO₂ ≥ 150 mmHg with a fraction of inspired oxygen (FiO₂) ≤ 40% and a positive end-expiratory pressure (PEEP) ≤ 8 cmH₂O]. Exclusion criteria included age < 18 years, decision to withdraw life support, hemodynamic instability with significant doses of vasopressors (dopamine or dobutamine > 10 μg/kg/min, epinephrine or norepinephrine > 0.5 mg/h), patient deeply comatose or sedated, and extreme temperatures (< 36 °C or > 39 °C). The modality of the first SBT was either a low-pressure support ventilation without PEEP or a T-piece, as per the recommendations [14, 15] and the usual care in the participating units. Criteria for SBT failure were respiratory rate ≥ 35 breaths/min or increase ≥ 50%, SpO₂ ≤ 90% or PaO₂ ≤ 50 mmHg (with FiO₂ ≥ 50%), heart rate ≥ 140 beats/min, new onset of supraventricular or ventricular arrhythmia, systolic arterial pressure > 180 or < 90 mmHg, alteration of consciousness, and diaphoresis or any signs of respiratory distress [14, 15]. Patients who failed the first SBT were included in the study.

This study was conducted in accordance with the amended Declaration of Helsinki. The protocol was approved by our institution’s local ethics committee (Comité de Protection des Personnes Ile-de-France IX, approval number 10-064). The protocol was considered a component of standard care and the patient’s consent was waived. Written and oral information about the study was given to patients or families.

No specific therapeutic interventions were recommended to the clinician after the failure of the first SBT. A second SBT, consisting of a 2-h T-piece trial [16], was performed within 24 h after the first SBT in all included patients. We chose the T-piece trial for the second SBT because this modality may be more prone to stress the cardiorespiratory system in order to diagnose WIPO [6]. Criteria for second SBT failure were the same as for the first SBT. Patients who succeeded the second SBT were extubated. Successful weaning was defined as patient alive and not reintubated within the 7 days following extubation, irrespective of the use of noninvasive ventilation [15]. We classified patients into three groups, according to the WIND classification [1], as follows: short weaning (successful weaning or death within 1 day after the first SBT), difficult weaning (successful weaning or death after more than 1 day but in less than 7 days after the first SBT), and prolonged weaning (successful weaning or death after 7 days following the first SBT).

Because there is no noninvasive consensual definition of WiPO, we considered three criteria proposed in the recent literature: (i) echocardiographic signs of increased left atrial pressure at the end of the SBT: E/A ratio > 0.95 and E/e’ ratio > 8.5 [17]; (ii) an increase of BNP (absolute change ≥ 48 ng/l) or NT-proBNP (absolute change ≥21 ng/l) concentration during the SBT [9]; (iii) an increase of protein concentration (relative change > 6%) during the SBT [8]. We further combined these criteria into two definitions of WiPO, as follows: a conservative definition (when at least two criteria were fulfilled) and a liberal definition (when only one criterion was fulfilled).

Transthoracic echocardiography was performed by a trained operator just before and at the end of the second SBT, as previously described [18]. Briefly, left ventricular ejection fraction was assessed using Simpson’s biplane method or estimated visually when poor spatial resolution did not allow sufficient identification of the endocardium. Wall motion analysis was also visually assessed [19]. In the apical four-chamber view, left atrial pressure were estimated by assessing early (E) and late (A) diastolic wave velocities at the mitral valve using the pulsed-wave Doppler; tissue Doppler early (e’) and late (a’) wave velocities were also assessed at the lateral mitral valve annulus. Pulsed-wave Doppler of the left ventricular outflow tract was assessed in the apical five-chamber view for cardiac output computation. The existence of significant (at least moderate) mitral regurgitation was assessed using color Doppler [20].

During second SBT, venous samples were collected to measure plasma levels of brain natriuretic peptide (BNP) or amino terminal pro-brain natriuretic peptide (NT-proBNP), protein, and high-sensitive cardiac troponin T (or troponin I in non-equipped centers) at three time points: before the SBT (in all patients), at the end of SBT (in all patients, whether it was a success or a failure), and 2 h after the start of SBT (only in patients reconnected prematurely to the ventilator because of SBT failure). We also assessed arterial blood gas analyses before and at the end of SBT.

ST segment measurements were recorded every minute during the second SBT using a continuous 12-lead electrocardiogram via the monitoring station. We adapted the third universal definition of myocardial infarction proposed by the European Society of Cardiology (ESC) [21] to define electrocardiographic WiCI, as follows: ST elevation in two contiguous leads (≥ 0.10 mV in all leads other than V₂–V₃; ≥ 0.20 mV in V₂–V₃ in men ≥ 40 years; ≥ 0.25 mV in V₂–V₃ in men < 40 years; ≥ 0.15 mV in V₂–V₃ in women), or ST depression ≥ 0.05 mV in two contiguous leads. As weaning shares some similarities with a cardiac stress test, another definition was extrapolated from the American Heart Association (AHA) statement for exercise testing as follows: ST elevation or depression ≥ 0.10 mV in two contiguous leads [22].

We hypothesized that the previously reported WiCI prevalences of 6–10% were probably underestimated given the ancient monitoring techniques used [23, 24]. We estimated that a sample size of 200 patients would allow detecting a prevalence of WiCI of at least 15% with an accuracy of 5% and an alpha risk of 5% (95% confidence interval). The data were analyzed using SPSS Base 20 (IBM-SPSS Inc., Chicago, IL, USA). Categorical variables were expressed as numbers [percentage], continuous data with normal distribution as means [standard deviation], and continuous data with non-normal distribution as medians [25th–75th percentiles]. We used the chi-squared or Fisher exact test to compare categorical variables between groups. The Kruskal-Wallis test was used to compare independent data with non-normal distribution and the Mann-Whitney test to compare paired data. Two-sided p values less than 0.05 were considered significant.

Among patients who failed a first SBT, weaning was short in 51/208 [25%], difficult in 95/208 [46%], and prolonged in 62/208 patients [30%] according to the WIND classification. No patients died in the short group while 9/95 patients [9.5%] and 35/62 patients [56.5%] died in the difficult and prolonged group, respectively. Patient characteristics were similar in the three groups except for a higher prevalence of heart failure with reduced ejection fraction, septic shock, and ventilator-associated pneumonia before weaning in the prolonged group, as compared to other groups (Additional file 1). Forty-one over 208 [19.7%] patients had a past history of coronary artery disease, and this prevalence did not differ between the three groups (Additional file 1). The clinical and biological parameters assessed just before the second SBT were also similar between groups, except for longer duration of mechanical ventilation since intubation, higher prevalence of the assist-control ventilation mode and of pulmonary consolidations, more fluid balance since admission, and lower values of MRC score and hemoglobin concentration in patients with poor weaning outcomes as compared to those with short weaning (Table 1). Patients with prolonged weaning had a longer length of stay in ICU and a higher mortality as compared to their counterparts (Additional file 1).

The median delay between the first and second SBT was 1 day [0–1]. One hundred and thirty-two over 208 patients (63.5%) failed the second SBT (Table 1). During the second SBT, NT-pro BNP and echocardiographic surrogates of left atrial pressure (E, E/A, and E/e’) increased in the failure group, but not in the success group, while an increase in protein concentration was observed in the two groups (Table 2). Only 17 patients had significant mitral regurgitation at the end of the SBT: 14 were mild and 3 were moderate. Overall, WiPO was diagnosed in 124/208 (59.6%) and 44/208 (21.2%) patients, according to the liberal and conservative definitions, respectively. WiPO was more frequent in patients who failed the second SBT as compared to successes, and whatever the definition used. WiPO did not influence the overall duration of weaning as assessed by the WIND definition (Table 3). The evolution of other cardiorespiratory parameters during the second SBT is reported in Additional file 2.

ST monitoring could be assessed in 177 patients during the second SBT, and WiCI was diagnosed in 36/177 (20.3%) and 12/177 (6.8%) of them, according to the ESC and AHA definitions, respectively. There were more patients admitted for cardiac arrest in the WiCI group as compared to their counterparts (Additional file 3). Changes in troponin were not different between patients with WiCI and their counterparts (Additional file 4). WiCI was not significantly associated with WiPO (Table 4, Fig. 1), neither with the outcome of the second SBT (Table 3). WiCI was more frequent in the prolonged weaning group when using the AHA, but not ESC definition (Table 3).

Two patients of the cohort were treated for an acute coronary syndrome after the second SBT, including one patient with a percutaneous reperfusion (the other patient presented a hemorrhagic shock which precluded any reperfusion therapy). Seven patients with a past medical history of coronary artery disease failed their second SBT and were explored with coronary angiography: two patients had coronary lesions without necessity of reperfusion therapy while five patients had no significant coronary lesions (Additional file 5).

WiPO is one of the main mechanisms of weaning failure [5]. We found a high prevalence of WiPO after a first SBT failure. This result is consistent with previous studies [8, 17, 25, 26]. The gold standard to evaluate LV filling pressures is to measure the pulmonary artery occlusion pressure using a pulmonary artery catheter. Because our patients were not equipped with such device, we chose to combine three noninvasive criteria as proposed by the recent literature, namely echocardiography [17], cardiac biomarkers [9], and protidemia [8]. We observed significant dynamic changes in these surrogates of left atrial pressure during SBT failure, and the prevalence of WiPO in our study (21 to 60%) is in accordance with that found with the gold standard pulmonary artery catheter (44%) [17].

Overall, WiPO was more frequent in patients who failed the second SBT as compared to successes, whatever the definition used (conservative or liberal). However, nearly half of the patients who succeeded the second SBT had at least one criterion for WiPO. This result is consistent with previous reports suggesting a significant increase in left atrial pressure during SBT, even in the case of success [27].

The lack of association between WiPO and weaning outcomes in our series may be explained by an early depletive management prompted by the initial evolution and the first SBT failure. In fact, more than half of patients received diuretics before the second SBT. Patient selection may also be important. Only one quarter of our patients had a short weaning (with a successful extubation less than 24 h after the first SBT), a proportion lower than that reported in the WIND study (57%). This discrepancy is at least in part explained by the fact that we included only patients who failed their first SBT. The incidence of WIPO in our study could be overestimated due to the increased proportion of unsuccessful SBT after a first SBT failure. The median delay between the first and the second SBT was very short in our study with a median of 1 day. These patients were successfully extubated on the next day (within 24 h after the first and the second SBT).

Prevalence of WiCI during the second SBT was less pronounced, as compared to WiPO, and varied between 6.8% and 20.3%, depending on the definition used. During weaning, silent myocardial ischemia was reported in high-risk postoperative patients [28] and in patients with a known coronary artery disease [23]. Prevalence of WiCI in the general population of critically ill patients varied between 0 and 70% across small cohorts [11, 24, 29], but the method of diagnosis and definitions often differed from ours. The association between WiCI and weaning outcome was weak in our study and only significant for the most restrictive definition derived from AHA statement for exercise testing. Electrocardiographic changes may not be specific to diagnose myocardial ischemia and may reflect myocardial injury whatever its cause [30]. However, ECG monitoring during cardiac stress is recommended as a first-line test to diagnose stable coronary artery disease and to detect transient myocardial ischemia [22, 31]. The association between WiCI and WiPO was also weak, suggesting that non-ischemic mechanisms may instead contribute to WiPO. In fact, the role of diastolic dysfunction seems central in the pathophysiology of WiPO [18]. The removal of positive-pressure support during weaning increases left ventricular preload and afterload and may impair left ventricular compliance [32, 33]. The ability of left ventricle to improve diastolic performance and maintain normal filling pressures during stress may be of paramount importance during weaning. Future strategies should aim at differentiating non-ischemic myocardial injury from myocardial infarction [30]. Biomarkers of cardiac injury (troponin) seemed of little help to detect transient myocardial ischemia during weaning in our study. Only three patients had significant variations in troponin I during weaning, and none of these patients experienced electrocardiographic changes. Further studies should assess the usefulness of other biomarkers with a shorter half-life time (e.g., myoglobin) in this setting.

Only two patients with significant modifications of ST monitoring were treated for an acute coronary syndrome after the second SBT. Although quite infrequent, aggressive therapeutic interventions (such as coronary reperfusion) may sometimes be required to successfully extubate those patients [34, 35]. The pre-existence of a known coronary artery disease (before admission to ICU) was not associated with WiCI. Whether undiagnosed coronary artery disease could be associated with WiCI remains unknown. Future studies should explore strategies aimed at detecting patients with clinically relevant cardiac ischemia during the weaning process.

Out study is the first to explore myocardial ischemia in a large cohort of patients who failed a first SBT. Strengths of our study include its prospective and multicentric design and the comprehensive assessment of cardiac function with multiple tools (echocardiography, continuous 12-lead ECG ST measurement, cardiac biomarkers). Limitations include the non-blinded and observational nature of the study, the absence of invasive monitoring which precluded any direct measurement of left atrial pressure, and the need to use multiple indirect criteria to define WiPO, owing to the lack of consensus in the literature. We did not assess lung ultrasonography because studies suggesting a role for this technique for the assessment of WIPO [36] were published after we started our study. In addition, the generalizability of our findings is limited by the fact that we examined a sub-sample of patients who had failed an initial SBT. Last, mitral regurgitation assessment was not exhaustive in our study and mainly used color Doppler, which may have led to quantification errors.