Discussion
MC is a potentially life-threatening complication in patients with MG, but the mortality rate has fallen dramatically over the past 60 years. The introduction of the neurological ICU has substantially improved early recognition of MC, identification of its precipitating factors and respiratory management of patients. The present work may help further improve the early recognition and care of patients who suffer MC by providing a picture of clinical characteristics (Additional file
1: Table S1) and even suggestions of baseline factors that may help predict survival.
Mean age at first MC onset requiring ICU management was 40.5 years. However, the median age at first MC onset was 55 years in a US study [
3]. One possible explanation for this discrepancy is ethnicity; other explanations include the differences in sample size, environmental factors and other population factors. In the present study, we found that first MC affecting people younger than 50 years affected women disproportionately, most of whom were aged 20–50; in contrast, first MC affecting people older than 50 did not show gender bias. These results are consistent with other studies [
3,
6,
8,
9]. The average interval from onset of MG to first MC requiring ICU management was 24.06 months in our cohort, much longer than the 8 months reported in another study [
3]. Our results are consistent with recent reports of a median interval from onset to crisis of 3 years [
9] and mean duration of MG prior to ICU admission of 3.8 years [
10]. A longer interval from MG onset to first MC probably reflects recent improvements in recognition of the disease, management of respiratory and bulbar conditions, and greater access to newer treatment modalities. Just over half our patients experienced their first MC within one year of symptom onset, consistent with a study showing that MC typically occurs within the first 2 years after MG diagnosis [
11].
While MG diagnosis in Europe and North America is most often supported using the tensilon test, the neostigmine test is used more often in China. In our study, the neostigmine test showed overall sensitivity of 88.0%, a little bit lower compared to the 96.8% reported by another study in China [
12]. These results validate the important role of this test for MG diagnosis in China. The proportion of patients in our cohort who took the repetitive stimulation test and gave a positive result was 83.00%, higher than the 77.4% reported in a cohort of 1108 Chinese MG patients [
12], and higher than the 75.9% reported in an Italian cohort [
13]. The higher rate of positive results on the repetitive stimulation test in our study may reflect the fact that we included all MC patients admitted to the ICU during the study period, none of whom had ocular MG. Sixty-two patients in 92 cases were positive for anti-AChR antibodies. This may underestimate the real prevalence of such antibodies, since this test is not routine in China because of resource limitations.
Infection, especially lower respiratory tract infection, was the most common identifiable precipitant of MC, followed by medication, and inadequate treatment/drug withdrawal. Other studies have also identified respiratory tract infection as the most frequent cause of MC, accounting for about half of crises resulting in ICU admission [
9,
10,
14]. Failure to comply with treatment or drug withdrawal was a frequent cause of MC accounting for 11 patients out of 113 in our study. Initial treatment with steroid led to exacerbation of MG in 30–50% of patients and decompensation in patients with MC, whereas 9–18% of them develop MC [
15]. In the present study, 7 patients out of 113 develop MC due to high-dose steroid therapy. Therefore, initiation of high-dose steroid should occur in a hospital setting, where the respiratory function can be monitored [
15]. Predictors of exacerbation from steroid include older age, bulbar symptoms, and lower score on Myasthenia Severity Scale [
15‐
17]. Our study showed pregnancy as a trigger of MC being responsible for crisis in 3 out of 113 patients, and study reported that pregnancy can aggravate MG in 33% of the MG cases [
5,
18,
19]. We suggest a detailed review of systems when the disease is getting worse, with attention to infectious sources, respiratory symptoms, and drug exposures (12). Physicians must pay careful attention to respiratory rate, difficulty with phonation, a quiet voice, weak neck muscles, work of breathing, and oxygenation. If the patient demonstrates vital capacity (VC) < 10–20 mL/kg or negative inspiratory force (NIF) < − 20 to − 30 cm H2O, diagnosis of MC is considered. However, these values are not derived from studies on patients with MG, but rather from studies in patients with GBS. We recommend that physicians should focus on the respiratory status of the patient, and trends in these symptoms, rather than relying on absolute numbers of VC or NIF. We also identified higher MG-ADL score at MC onset as a potential indicator that ICU care will be needed. Indeed, MG-ADL score > 18 points at MC has been reported to predict the need for ICU management with 75% sensitivity and 77.8% specificity [
14]. Surprisingly, we detected severe hypercarbia in our cohort before intubation (mean PCO
2, 48.78 mmHg). Since MG-ADL score at MC onset correlated positively with PCO
2 before intubation, respiratory status may be tightly associated with MC symptoms, and hypercarbia may affect daily activities of patients with MC. Mean duration of ICU stay was 12.3 days in our study, similar to the median of 14 days reported in a US study thirty years ago [
6] or the median of 13 days reported in a US study more recently [
3]. The mean duration of ventilation of 190.3 h in our study is similar to the 8 days reported in a US study [
20]. But a study from India showed that the median duration of ventilator was 14 days in a group of patients with MC [
21], this result was similar to the previously reported duration of 13 days [
22]. We found that pre-intubation PCO
2 and MG-ADL score at MC onset were associated with duration of ventilation and ICU stay. Higher PCO
2 prior to mechanical ventilation may indicate more severe condition that will likely require extensive respiratory support and ICU management. Respiratory management is important for MC, one study showed that bilevel positive airway pressure (BiPAP) before the development of hypercapnia was useful in preventing intubation and prolonged ventilation [
23]. Another study revealed that hypercapnia at onset predict BiPAP failure and subsequent intubation [
24]. Therefore, high PCO
2 level before intubation may be an independent predictor of prolonged intubation. Thomas et al. identified three risk factors were significantly associated with prolonged intubation, including pre-intubation serum bicarbonate > 30 mg/dl and age > 50 years [
3]. In our cohort, the proportion of patients remaining intubated for longer than 2 weeks was 42.9% (21/49) with both hypercapnia and age > 50 years. However, designed prospective study is required verify the statistical significance of various parameters leading to prolonged intubation.
Over 80 % of our patients showed good functional outcome during follow-up. This likely reflects the potentially reversible character of MG and substantial advances in therapeutic and supportive measures [
11]. Study has showed that MG is often associated with better functional outcomes at one year than other diseases requiring neurocritical care [
25]. However, patients with intermediate and poor outcome had older age of first MC onset, and lower pH and PO
2, as well as higher PCO
2 before intubation. Previous study retrospectively included 38 MC patients admitted to the Neuro-medical ICU, and found that 4 patients died in hospital, and the remainder of patients with different age of MC onset (older (> 50 years) and younger (< 50 years) patients) did not show differences in long-term outcome [
10]. However, one study found that being older than 50 at first MC independently predicted prolonged intubation [
3], while another reported that being younger than 40 at MG onset was associated with higher likelihood of remission [
13]. The associations between age of first MC onset and outcomes need to be clarified in larger studies with long follow-up. A study showed that pre-ventilation pH below 7.30 and high PCO
2 were associated with poor functional outcome and death [
26]. A study comparing MC patients in the ward or in the ICU reported that only those in the ICU had abnormal arterial blood gases, and that patients in the ICU had lower pH and higher PCO
2 [
14]. Low pH and high PCO
2 indicate chronic respiratory acidosis, which may be associated with severe disability and death, especially in MG patients who experience MC. Little is known about the outcome of first MC patients suffering from acute severe exacerbations following ICU discharge. In the present study, pre-intubation PCO
2 and age of first MC onset were considered to be factors associated with survival. Therefore, in MC patients with extremely high PCO
2 level before intubation may obtain poorer prognosis, especially in patients with older age.
By the end of follow-up, 21 of 113 patients in our cohort had died (18.6%).This mortality rate is near the high end of the range of 6–30% reported for MC patients in several studies [
3,
9,
21,
27,
28]. In a Chinese cohort from Hong Kong, 35 MG patients experienced crisis and 2 died (5.7%) [
29], but in a cohort from India, mortality was in 3 out 10 (30%) during MC (30%) [
9]. However, the mortality rate of MC fell from 42% in the early 1960s to contemporary rates of 4 to 10% with the improvement of the advent of IVIg and plasma exchange and ICU management [
3,
6]. The relatively high mortality rate in our study may reflect the fact that we included all consecutive patients who presented in the neurological ICU during the study period. MC patients were managed in the general ward were not included in the present study. Study has shown that compared to MC patients who received general ward management, MC patients with ICU management had higher MG-ADL scale scores and higher MGFA classification [
14]. There could be selection bias, since more seriously ill patients could be selected in the present study. The other fact needed to consider is the ground clinical reality in developing countries, and poor awareness on this part of patients. In addition, drug nonaffordability is the actual reality in China. For example, both plasma exchange and IVIG are not covered by the medical insurance, and the expense on plasma exchange/IVIG is more than the annual income for some Chinese family. In resource-challenged settings like China, vigorous and concerted efforts should be made in MC prevention, timely identification, emergency intervention, and aggressive treatment.
Our study has several shortcomings. First, study have shown that chronic obstructive pulmonary disease (COPD), diabetes mellitus, atrial fibrillation, hyperlipidemia, myocardial infarction, and malignant tumors, were highly associated with death in the MG population [
30]. Another study found that at least one comorbid disease was diagnosed 93% patients with late-onset MG (after 60 years) [
31]. Our results showed that there was no significant difference among patients with different outcomes regarding the comorbidities, probably due to the small sample size and relatively short follow-up time. Second, our population may have been affected by referral bias because our hospital is a tertiary referral center. With regard to hospital-related factors, it is possible that the apparent benefit of neurological care may be far different from other centers. Some management artefact and treatment strategies may lead to changes in outcomes. As a result of the retrospective design of our study, we may have failed to include certain patients who were not entered properly in the hospital computer system. Third, some patients were unable to receive IVIG or plasma exchange due to the family’s economic hardship. The usage of immunosuppressants in the treatment of MG has greatly changed the outcome of MG patients [
32]. One study indicated that azathioprine therapy independently predicted good clinical outcome of MG patients [
29]. Another study showed that combined prednisolone-azathioprine treatment reduced the proportion of recurrent MCs, and the number of mechanical ventilation events and ICU admissions were also reduced [
33]. Furthermore, the administration of immunosuppressants was found to be closely associated with a decreased risk for death of MG patients [
29,
30]. However, only 12 patients received immunosuppressants in the present study. These deficiency in treatment may be associated with the poor prognosis in same patients. Finally, we did not analyze data related to other parameters that might have affected clinical outcomes, including maximal expiratory pressure and maximal inspiratory pressure on pulmonary function tests.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.