Introduction

Myasthenia gravis is a rare auto-immune disease affecting the neuromuscular junction characterized by fluctuating skeletal muscle fatigability potentially that can potentially lead to paralysis1. Severity of exacerbation is marked by swallowing disorders which may lead to aspiration pneumonia and/or respiratory muscle paresis which may require invasive mechanical ventilation. Respiratory failure requiring mechanical ventilation caused by myasthenia gravis defines myasthenic crisis (MC). MC occurs in 15–20% of patients with myasthenia gravis and is treated by mechanical ventilation, acetyl-cholinesterase inhibitors, immunosuppressants such as corticosteroids and specific immune-modulatory therapeutics including plasma exchange, polyvalent intravenous immunoglobins or monoclonal antibodies2.

Weaning failure during MC occurs in 27–60% of cases3,4,5,6, higher than in the general ICU population7 suggesting that criteria described in weaning guidelines might be insufficient to identify this at-risk population8,9. Standard clinical criteria and usual weaning trials might not be sensitive enough to predict weaning failure in MC. Within a general intensive care unit (ICU) population, maximum expiratory pressure, a surrogate for cough strength, may predict weaning failure10,11. Thomas et al. previously showed that pulmonary function testing (PFT) could identify MC patients at risk of prolonged ventilation12. However, using pulmonary function testing to predict weaning failure has never been assessed. Moreover, in the general ICU population, prolonging spontaneous breathing trials do not reduce the risk of weaning failure13. On the other hand restrictive extubation criteria might prolong weaning from MV.

In this study we sought to determine the incidence of weaning failure after the 1st extubation attempt in a cohort of patients with MC in a specialized, highly experienced intensive care unit, where prolonged SBT and repeated PFT were performed at the bedside. Secondary objectives were to assess risk factors and outcome associated with weaning failure after first extubation in MC as surrogates for safety and efficiency. Since conditioning extubation on the success of prolonged SBT and PFT might lengthen the weaning period, we also compared the characteristics of patients with prolonged weaning.

Methods

We undertook a monocentric retrospective study. We included patients suffering from myasthenic crisis between January 2001 and January 2018. Patients identified as suffering from myasthenia gravis (G700 or G730), intubated and ventilated (GLLD004, GLLD006, GLLD008, GLLD015) in the ICU were retrospectively identified using their International Classification of Diseases 10 codes. Diagnosis of MG relied on the occurrence of clinical symptoms of myasthenia gravis, confirmed by specific tests (antibody testing or pathological decrement on repetitive nerve stimulation or improvement after cholinergic medication). We excluded patients with congenital myasthenia, Lambert-Eton syndrome, those bearing a tracheostomy at the time of admission as well as patients ventilated for less than 24 h.

We collected demographic characteristics including age, gender, height, weight, medical history of cardiac failure, or other respiratory diseases. We also recorded whether the myasthenic crisis was inaugural, presence of post synapse nerve failure identified by repetitive nerve stimulation, acetylcholine receptor (AChR) or anti-Muscle Specific Kinase (MuSK) antibodies, a history of thymoma, a history of thymectomy as well as usual treatments (corticosteroids, mycophenolate mofetil, azathioprine) and dose of acetylcholinesterase inhibitors. We also collected the Sequential Organ Failure Assessment (SOFA) score at the time of admission, the ventilator mode, as well as the occurrence of ventilator associated pneumonia, septic shock or acute respiratory distress syndrome.

Trigger factors of the myasthenic crisis were identified: infectious (pneumonia, infection of the upper airway, other causes), surgical (thymectomy, other surgery), change of medication (introduction of corticosteroids, lowering of corticosteroids, and introduction of a contraindicated drug) and other. Clinical characteristics of the myasthenic crisis as well as severity were assessed using the myasthenia muscle score14, which quantifies muscle strength and endurance of the eyes, bulbar symptoms, limbs and axial tonus. The MM score ranges from 0 to 100, with lower scores denoting less muscle strength and endurance. The head lift test was part of the MM score and consisted in the ability to lift the head against resistance. Specific treatments administered for myasthenic crisis were recorded, including intravenous immunoglobulins, plasma exchange or increased dosage of corticosteroids. Symptomatic treatment included increasing the dose of acetylcholinesterase inhibitors to the maximum permitted dosage.

The use of non-invasive mechanical ventilation (barring short-term non-invasive ventilation (NIV) used for pre-oxygenation before intubation) before instauration of invasive MV was collected. Criteria for intubation were based on signs of respiratory distress (Respiratory Rate > 30/min), respiratory acidosis (pH < 7.36 and PaCO2 > 45 mmHg) or clinical evidence of hypercapnia or hypoxemia. Patients were ventilated using a volume assist controlled mode aiming at a 8 mL/Kg tidal volume, a maximum sensibility inspiratory trigger and a positive end expiratory pressure (PEEP) at a minimum of 5 cm H2O, and the lowest FiO2 required to achieve blood oxygen saturation targets, in the absence of hypoxemia and acute respiratory distress syndrome. The date of intubation, length of mechanical ventilation and length of weaning were recorded.

At the time of admission and throughout the period of mechanical ventilation, pulmonary function testing was performed by experienced technicians twice a week. Pulmonary function testing was performed with the patient sitting in bed. Recorded parameters included vital capacity expressed as percent of theoretical value (VC%), maximal inspiratory and expiratory pressure (MIP and MEP) using a spirometer (Spirodoc®, Medical International Research, Rome, Italy) and a respiratory pressure meter (MicroRPM®, CareFusion, San Diego, CA, USA) connected to the endotracheal tube. The best value out of three reproducible values was recorded. Variations of vital capacity during SBT were measured by comparing pulmonary function testing immediately before SBT and 4 h after initiation of SBT.

SBT were considered when FiO2 was 50% or less, PEEP 8 cmH2O or less, when the patient was no longer receiving vasopressors or was receiving low doses of vasopressors, when the trigger of myasthenic crisis was considered resolved, when the signs and symptoms of myasthenia gravis were controlled as assessed by the MM score and a vital capacity above 30% of predicted value. Spontaneous breathing trials using a T-tube were initiated over increasingly long periods until the patient was able to breathe spontaneously over an 8-h period.

Physiological parameters (systolic, diastolic and mean blood pressure, respiratory rate, blood oxygen saturation and the flow of oxygen required to maintain said oxygen saturation) were continuously collected throughout the spontaneous breathing trial. Spirometry and maximal respiratory pressure were collected before initiation of the SBT and 4 h thereafter. Finally, arterial blood gases were obtained at the end of the 8-h SBT. Criteria for passing the SBT included adequate subject comfort defined as the absence of severe dyspnea, request for SBT interruption or agitation and respiratory pattern, without polypnea above 30/min, hemodynamic stability defined as variations of < 30% relative variation of blood pressure, heart rate, respiratory rate during SBT and < 30% relative variation of vital capacity between the initial measure and the measure after four hours of SBT, without hypercapnic acidosis (pH < 7.36 and pCO2 > 6 kPa at the end of the SBT. Following a successful SBT, mechanical ventilation was systematically re-introduced, enteral nutrition was interrupted and the patient was extubated the following morning. Intensive respiratory physiotherapy was provided at least twice daily. Liquid and solid food was carefully reintroduced. Only the first extubation attempt was considered for this study. Extubation failure was defined as death or the need to reinstitute mechanical ventilation within 7 days as previously described. Successful extubation was defined as the absence of the need reinstitute mechanical ventilation within the 7 days after the 1st separation attempt. Prolonged weaning was defined as the requirement of more than 7 days between the first SBT and the first extubation attempt7.

Ethics

The study was performed in compliance with the ethical principles formulated in the declaration of Helsinki and was approved with a waiver for informed consent by the French Society of Intensive Care (SRLF) ethical review board advice N°19-56.

Statistical analysis

Patients with missing data regarding the weaning period were excluded from the study. For each variable, the proportion of missing data is provided. Quantitative datas are presented as median with interquartile ranges or mean and standard. The main outcome is shown as the percentage of weaning failure after the 1st extubation attempt in patients receiving MV for MC. Characteristics of patients successfully weaned after first extubation were compared to unsuccessfully extubated patients using nonparametric tests. No model was evaluated for the main outcome because of the paucity of predictors for weaning failure and the low number of event.

Concerning the prediction of prolonged weaning, values are expressed as nonparametric variables and compared using a Wilcoxon test. For all significant variables in univariate analysis on the original data (p < 0.05) missing values were estimated. Logistic model for dichotomous variables and regression for quantitative variables. Five replications were made to taken for the missing pooled data. Statistical analyses were done with R statistics 4.2.1. Values less than 0.05 were deemed statistically significant.

Results

Incidence, risk factors and outcome of weaning failure after 1st extubation

18/126 (14.3%) patients suffered from the main outcome of extubation failure upon 1st attempt. Three unplanned extubations occurred, requiring in all cases swift reintubation. These accidental extubations were not considered separation failures (Fig. 1).

Figure 1
figure 1

Flow chart of the study. Mo, months.

Full size image

Characteristics of patients at the time of admission as well as the characteristics of the disease were similar when comparing weaning failure to success after 1st attempt (Table 1). Successful weaning was more frequent when a precipitating factor for MC was identified (86/108 (79.6%) vs. 8/18 (44.4%); p = 0.004, (Table 1)). During SBT, vital parameters and results of blood gas analysis were similar in both groups (Supplementary Fig. 1). PFT could not help predict the extubation failure upon 1st attempt as MM, VC%, MIP or MEP were similar between groups at any time during their stay (Fig. 2). Only inability to lift the head (Table 2, p < 0.001) was associated with increased extubation failure.

Table 1 Patients’ characteristics in the overall cohort and a compared between successful weaning after 1st extubation or not.
Full size table
Figure 2
figure 2

Evolution of Myasthenia gravis control during ICU stay according to Myasthenia muscular score and pulmonary functional tests in patients with successful weaning or not after 1st extubation. VC, vital capacity expressed in % of theory; MIP, maximal inspiratory pressure in cm H2O; MEP, maximal expiratory pressure in cm H2O.

Full size image
Table 2 Pulmonary function tests from intubation to the end of spontaneous breathing trial compared between successful weaning or not after 1st extubation.
Full size table

105 out of 126 (83.3%) patients underwent an 8-h SBT without showing any pre-established criteria for non-extubation. 17 patients were extubated despite a SBT duration of less than 8 h, and 5 despite a decrease of more than 30% in vital capacity [14(13%) vs. 3(17%)), and 4(4%) and 1(8%)] respectively in the weaning success and failure groups.

When analyzed case by case (Table 3), NIV was associated with successful weaning after 1st attempt in 9/75 (12.0%) cases and in 4/12 (33.3%) cases of weaning failures p = 0.051. Extubation failures after 1st attempt were attributed to an insufficient control of myasthenia gravis (11/18, 61%) manifesting as aspiration in 6/11 (55%) cases, whereas other causes were non-specific to myasthenia gravis (7/18, 39%) were considered plurifactorial (4/18 patients), due to laryngeal edema (2 patients) and one cause was unknown.

Table 3 Case by case analysis of patients with weaning failure.
Full size table

Incidence risk factors and outcome associated with prolonged weaning

56/116 (48.3%) patients were extubated after a prolonged period of weaning. Patients’ characteristics were similar in both weaning duration groups (Table 4).

Table 4 Baseline Patient Characteristics according to weaning duration.
Full size table

In univariate analysis, SAPS II (45 [29, 52] vs. 53 [38, 56], p = 0.045) use of both intravenous immunoglobulins (IVIG) and plasma exchanges (PE; 2 (4.0%) vs. 14 (29.2%), p = 0.002), ventilator acquired pneumonia (10 (17.5%) vs. 20 (36.4%), p = 0.042) ICU adverse event (6 (10.5%) vs.19 (34.5%), p = 0.005) were associated with weaning duration. MM score did not differ between groups at the time of admission (33 [23–56] vs. 39 [25–58]; p = 0.525), at the time of intubation or SBT (46 [37–57] vs. 46 [38–56]; p = 0.846) (Table 5). PFT could help differentiate patients with prolonged weaning as VC% at the time of intubation (36 [28–51] vs. 28 [22–36]; p = 0.009) and the minimal VC% (30 [21–39] vs. 23 [19–29]; p = 0.023) were associated with prolonged weaning of MV (Table 5). However VC% at the beginning or during SBT did not significantly differ. MIP at the time of admission was associated with median weaning duration (30 [20–36] vs. 19 [10–24] cmH2O; p = 0.001). MIP at the time of intubation was associated with median weaning duration (30 [20–35] vs 19 [12–22] cmH2O; p = 0.001).

Table 5 Myasthenia muscle score and pulmonary functional tests in patients according to weaning duration. SBT Spontaneous breathing trial.
Full size table

Prolonged weaning was not associated with less episodes of extubation failure (8/60 vs. 10/56; p = 0.501). Length of stay in the ICU was 17 [13, 29] days vs. 29 [18, 39] days (p < 0.001) while length of stay in the hospital was 34 [23, 51] days vs. 48 [33, 71] days (p = 0.005). Mortality was low in both groups (2(3.6%) vs. 1(1.8%)).

In multivariate analysis, for each additional day of VM, the risk of prolonged weaning was increased by 16.4% (95% CI: 9.2–24.1%; p < 0.0001), the presence of upper respiratory tract infections was found to be a positive factor reducing risk by 7.2% (95% CI: − 57.8% − 0.9%; p = 0.013), and for each additional H2O cm of MIP, the risk of prolonged weaning was reduced by 3.4% (95% CI: − 6.5% − 3.0%; p = 0.035). The Area under the ROC curve of the model was 0.753 (p < 0.001).

Discussion

The main result of this study is to show a rate of extubation failure after 1st attempt of 14.3% in a cohort of 126 patients admitted for myasthenic crisis using a prolonged SBT protocol with PFT in a referral center. Extubation failure occurred more often in patients with no identified trigger for MC or those with an inability to lift their head whereas PFT could not help identify patients with extubation failure. Causes of extubation failure were equally attributed to insufficient myasthenic control and non-myasthenia gravis related causes. Median MV duration was 14 days and identification of a trigger factor, MIP and MV duration prior to weaning were identified as risk factors for prolonged weaning.

Liberation failure rates in myasthenic crisis range from 27 to 64%3,4,5,6,15. In this cohort we report a unexpectedly low rate of extubation failure at 14.3%. This difference may be related to a center effect or patient selection biasrather than a specific weaning protocol. Compared to previous studies, patients in our cohort were older, but the use of PE or IGIV was similar to that in Rabinstein et al.and the MV total duration was similar to other cohorts3,4,5,12. Despite similar demographic datas, selection bias cannot be ruled out without randomization.

The incidence of ventilator acquired pneumonia might have decreased over time through implementation of prevention bundles which might account for our results. If we consider the incidence of ventilator associated pneumonia as a marker of general care, we report a 26.9% incidence on day 14 which is comparable to data collected in other ICU and almost twice as low as in other MC cohorts3,6,12,16. Thus, the reduced incidence of VAP may explain some of our findings.

Also, there have been significant improvements in the management of patients with myasthenia gravis, particularly regarding baseline treatment. However, the treatments for crises that remain the standard today are IVIG and PE, both used equally across our study cohort. Thus, patients might have benefited from the overall improvement in intensive care during the period of our study; nevertheless, there is no trend towards a decrease or an increase in extubation failures over the study period that were distributed equally during the two halves of the study period.

An accurate prediction of patients at risk of weaning failure after extubation failure is difficult8. In this cohort, patients were followed longitudinally, and extubation was decided based on general and myasthenia gravis related clinical parameters, as well as respiratory function testing. Patients were then extubated when the prolonged SBT was conducted, and the values of functional pulmonary evaluation were satisfactory. In a cohort study, Seneviratne et al.3 identified that in myasthenic crisis, weaning failure was associated with lower forced VC. We did not reproduced these results. In our cohort, the MM score, VC%, MIP, MEP improved at the time of extubation. As PFT were a criterion for extubation, it is not surprising that these were not identified as risk factors of extubation failure3.

However, we still observed a number of weaning failure after first extubation despite prolonged SBT and PFT normalization. Axial deficit is a potential marker of weaning failure. It could be associated with a patient's inability to reposition in bed, weaker cough leading to respiratory exhaustion, which have been previously described in general ICU8.

When analyzed individually, 60% of extubation failures was attributed to incomplete control of myasthenia gravis, leading to respiratory exhaustion and/or aspiration, the other half to laryngeal edema or respiratory exhaustion related to other factors. Second, determining the risk of aspiration is notoriously difficult in patients with endotracheal tube. Swallowing was assessed subjectively through endotracheal cuff deflation and saliva deglutition. Gag reflex may help predict aspiration and be indicative of insufficient myasthenic control and could be further evaluated19.

The use of non-invasive ventilation (NIV) was not protocolized and not used systematically in this cohort due to the risk of swallowing disorders associated with myasthenia gravis. NIV was used to treat respiratory failure after extubation but not as a prophylactic technique to reduce the risk of weaning failure. Among the 13 patients who received NIV after extubation, 9 were successfully weaned, and 4 required MV reinstatement. Its use as a bridge to extubation could represent a promising technique to limit the risk of weaning failure and warrants further evaluation.

In our cohort of 126 patients with MC, we observed a low rate of weaning failure after the first extubation, with only 18 events. Although our study is one of the largest evaluating MV weaning in patients with MC, the low event rate precludes us from performing a comprehensive analysis of the factors associated with failure, representing a limitation of our study. A multicenter design could help increase the power of such a study and enhance its external validity.

Some patients might be able to be successfully weaned earlier, before reaching prolonged SBT and normal PFT. In our cohort 96.7% of our patients underwent a 8 h SBT. Some patients were thus extubated following a shorter SBT when the clinical staff deemed the chance of success acceptable. Conditioning extubation on successful prolonged SBT and PFT normalization seems relevant since myasthenia gravis is characterized by fluctuating neuromuscular impairment, but the benefit of prolonged MV to avoid extubation failure has to be balanced with the risk of weaning failure which increases ICU length, costs and mortality8,17. We therefore worried that conditioning extubation to successful prolonged SBT and normal PFT would lengthen the duration of MV and morbidity. The observed duration of MV was 14 days in our cohort compared of 9–20 days in other cohorts, suggesting that applying such a procedure does not dramatically increase the duration of mechanical ventilation3,4,5,12. Additionally concerning safety, the mortality rate in our cohort (4%) was not higher than in other cohorts, which in some cases reached 10%6,12,18.

The main risk factors for prolonged weaning were MIP, MV duration and upper respiratory tract infection. This is in accordance with previous findings, which had identified a VC alteration after intubation as a risk factor for prolonged ventilation12. Of note, our weaning protocol did not include a MIP cut-off value to start the weaning process, since VC and MIP both indicate diaphragmatic function which is impaired during MC. It is thus questionable whether MIP was a risk factor, or a feature taken into account to prolong MV in a self-fulfilling prophecy.

The causality between a specific weaning procedure on weaning failure and its effect on MV duration remains elusive barring a randomized controlled trial. Such a trial is unlikely to be undertaken, due to the rarity of the disease. The reproduction of our findings in other centers implementing our procedure might strengthen the plausibility of a causal link.

Conclusion

In this monocenter study in a tertiary care specialized center extubation failure after first attempt in myasthenic crisis was 14.3%. Extubation failure occurred more often in patients with no trigger for myasthenic crisis or those unable to lift their head. Most failures were attributed to an insufficient stabilization of myasthenia gravis and aspiration. Whether prolonged SBT and PFT reduces weaning failure after first extubation without substantially prolonging duration of mechanical ventilation warrants further evaluation.