Myasthenia gravis (MG) is a typical autoimmune disease characterized by fluctuating muscle weakness. Over a 5-year period, thymectomy has significantly improved clinical outcomes and reduced prednisone requirements (1). Despite advances in medical treatment and surgical approaches, 3%−30% of patients with MG still experience postoperative myasthenic crisis (POMC) (2). POMC, the most severe complication of thymectomy, is characterized by a rapid worsening of symptoms, acute respiratory failure requiring mechanical ventilation, and even death. Several risk factors predicting POMC have been identified, including disease duration < 3 months, bulbar symptoms, vital capacity < 80%, preoperative myasthenic crisis (MC) history, and preoperative Osserman stages (IIb, III, and VI) (2, 3). In multiple autoimmunity diseases, the biomarkers from routine blood tests have been shown to be closely associated with the disease process (4–6). However, an easily accessible biomarker for POMC has not yet been proposed.

Bilirubin, generally considered the end product of heme metabolism, has potent anti-oxidative and anti-inflammatory effects (7). Indirect bilirubin (IBIL), also known as unconjugated bilirubin, originates mainly from the breakdown of red blood cells and is converted into direct bilirubin (DBIL) by UGT1A1 in the liver. Recent studies have shown that bilirubin could have opposite effects on human diseases, including coronary heart disease, cancer cachexia, and chronic obstructive pulmonary disease, depending on different bilirubin concentrations (8–10). Similarly, Zhang et al. (11) reported that low and high concentrations of bilirubin have opposite effects on nicotinic acetylcholine receptor (nAChR) function. However, the association between bilirubin levels and MG severity remains unclear. Yang et al. (12) revealed that low bilirubin levels may be a risk factor for MG aggravation, but Misra et al. (13) reported that high bilirubin levels are a biomarker of poor outcome in patients with MG. Bilirubin assays are routine, easily available preoperative examinations; therefore, bilirubin levels may have great clinical application as a biomarker for the prediction of POMC occurrence.

In this retrospective cohort study, we analyzed the dose-response association between preoperative bilirubin subtype levels and POMC incidence. In addition, we aimed to establish a threshold for bilirubin subtype levels to predict POMC occurrence.

This is the first study to investigate the association between preoperative individual bilirubin subtype levels and POMC incidence. Preoperative IBIL levels were higher in the POMC group than in the Non-POMC group, and a linear dose-response relationship was shown between the IBIL levels and POMC incidence. Interestingly, the IBIL level, especially that exceeding the normal upper limit, was an independent risk factor for POMC progression after controlling for the classic clinical risk factors. Therefore, high preoperative IBIL levels may serve as a routinely available biomarker for POMC incidence.

POMC is the most severe complication of thymectomy, and patients need mechanical ventilation and intensive care unit care. Thus, the early identification of risk factors is critical for reducing POMC occurrence. Lung function, bulbar symptoms, MGFA classification, and disease duration are classic clinical risk factors, but the proportion of patients with a disease duration < 3 months was not significantly different between the Non-POMC and POMC groups in our study. Moreover, a previous report showed that a history of MC was also a risk factor for POMC (2), which is also consistent with our findings. Although the RR-AChR antibody, miR-150-5p, and complement levels have emerged as potential biomarkers to predict MG progression (17, 18), convenient and inexpensive biomarkers are urgently needed.

Bilirubin was previously believed to have potent antioxidant and protective effects against autoimmune diseases. However, the correlation between bilirubin levels and MG severity remains inconsistent. Yang et al. (12) showed that serum IBIL and DBIL levels were lower in patients with MG than in healthy controls. IBIL and DBIL levels were correlated with disease severity and MGFA classification. Misra et al. found that serum bilirubin levels were increased in MG patients with poor outcomes (MGFA classifications III, IV, and crisis). Although this retrospective study had a longer follow-up period (1991–2016), only 81 patients' data were included (13). In our study, preoperative IBIL levels were significantly higher in the POMC group than in the Non-POMC group. In RCS regression analysis, line-response relationships were found in both TBIL and IBIL with POMC after adjusting for clinical variables. However, a U-shaped relationship was observed between DBIL and POMC in the adjusted model. Because IBIL accounts for the majority of TBIL, the relationship between TBIL and POMC incidence may be attributed to increased IBIL levels. Compared with the first quartile, the second quartile of IBIL level was negatively correlated with POMC incidence in the unadjusted model (Figure 2A). Hence, the risk threshold of the IBIL level is important in POMC progression for clinical judgment.

Recently, bilirubin has been found to have both anti-oxidant and pro-oxidant effects at different concentrations. Mild hyperbilirubinemia could protect against cellular reactive oxygen species (ROS) damage and have a beneficial effect in autoimmune diseases, including systemic lupus erythematosus, diabetes, and inflammatory bowel disease (IBD) (6, 19, 20). However, severely high bilirubin levels can cause permanent neurological impairments in neonates. Peng et al. found that DBIL dose-dependently predicted poor outcomes in patients with acute ischemic stroke after thrombolysis (21). When the total bilirubin level exceeded 13 μmol/L, the risk of COPD and CHD increases (8, 10). Hence, the beneficial or harmful effect on normal physical function or disease may depend on the IBIL concentration, but there is no certain threshold.

In vitro experiments, when the IBIL level exceeds 3.6 μmol/L, especially 15 μmol/L, Bianco et al. (16) reported that IBIL has significant pro-oxidant and cytotoxic effects. In our study, all IBIL data were obtained from our hospital, and the normal range for IBIL was 0–14 μmol/L. The starting range of severe cytotoxic effects of IBIL is approximately equal to the upper limit of its normal range, indicating that exceeding the upper limit of IBIL is harmful to immune homeostasis. Interestingly, high IBIL levels (>14 μM) were significantly related to an increased incidence rate of POMC. Compared with the Youden index, the upper limit value of the IBIL level was easily obtained from clinical data and had more clinical judgment value for POMC prediction. Therefore, with 14 μmol/L as the cut-off value, the patients with MG were divided into high and low IBIL subgroups. By controlling for clinical risk factors, a high IBIL level was an independent predictive biomarker for POMC. In patients with acute ischemic stroke after thrombolysis, increased DBIL before thrombolysis has a greater incremental predictive value for poor outcomes (21). Similarly, based on the IDI and NRI analyses, we found that predictive ability can be significantly improved by including the IBIL variable in the model.

IBIL is covered by DBIL in the liver. MG is also a systemic autoimmune inflammatory disease that can accompany non-motor symptoms, including headaches, sleep disturbances, and even taste disorders (22). In this study, patients with chronic hepatitis and abnormal ALT or AST levels were excluded. Although liver injury was reported in MG, few liver functions were observed in patients with non-motor symptoms. Immune-erythroid cells collected from bone marrow can promote the CD11b⁺ human peripheral blood mononuclear cells to produce proinflammatory cytokines (23). Shi et al. (24) found that the bone marrow hematopoietic stem could increase the number of monocytes and neutrophils, exacerbating central nervous system inflammatory injury in multiple sclerosis. These findings indicate that red blood cells may have immunomodulatory roles in autoimmune diseases, including MG. Hence, we could reasonably speculate that the imbalance of immunity of MG influences red blood cell and liver function, leading to elevated IBIL levels.

High IBIL levels are associated with a higher risk of POMC, and IBIL levels could promote MG progression via direct or indirect mechanisms. Zhang et al. found that IBIL can directly affect AChR function in superior cervical ganglion. Low IBIL concentrations could increase the peak amplitude of nicotinic AChR channel currents, but high concentrations of IBIL (3, 4, or 5 mM) could inhibit the function of AChR (11). The AChR subtype in the superior cervical ganglion is different from that of the neuromuscular junction. Whether high concentrations of IBIL directly affect AChR function needs to be explored in electrophysiological assays. MG is a typical T cell-dependent, B cell-mediated autoimmune disease. High IBIL levels can induce cells to produce excess ROS, which is considered an important bridge between innate and adaptive immunity (25). Mitochondrial ROS contribute to the maintenance of CD4 and CD8 T cells. Increased mitochondrial O₂^.− in T cells show higher levels of interleukin (IL)-4, IL-16, IL-17, tumor necrosis factor α, and inferno γ (26). Experimental autoimmune MG rats with diabetes experienced more severe clinical symptoms and increased Tfh, Th17, and memory B cell populations (27). This may be because high glucose intake induces the formation of mitochondrial-derived ROS, leading to the differentiation of Th17 cells through tumor growth factor-β activation (28). Additionally, ROS levels are much higher in activated B cells, which is essential for B cell proliferation (29). Gilljam et al. (30) revealed that ROS could play an important role in sustaining high immunoglobulin production when TLR9 induces B activation. ROS, as secondary messengers, can directly or indirectly act on various transcription factors, such as PKC/NF-κB, PI3K/AKT/mTOR, and RAS/ERK/AP-1 signaling pathways to mediate B cell and T cell differentiation (29, 31). The aforementioned findings suggest that ROS could induce T cell and B cell activation and promote autoimmune disease progression. Immune cell metabolism is central to inducing B and T cell differences; however, the characterization of metabolism in MG is still unknown. This study showed that high IBIL levels aggravate MG symptoms, which may depend mainly on the production of large amounts of ROS.

Recently, bilirubin nanoparticles have shown therapeutic efficacy in cardiac ischemia or reperfusion injury, rheumatoid arthritis, and IBD mouse models (32–34). These effects are mainly due to potent ROS scavenging, endogenous antioxidant molecules, and anti-inflammatory properties (35). Herein, when the IBIL level exceeded the normal upper limit, we found that the adverse effect of bilirubin was apparent in the progression of MG. Therefore, it is reasonable to speculate that bilirubin nanoparticles may not reach therapeutic efficacy when bilirubin levels are abnormally high in the body. Furthermore, bilirubin nanoparticles may cause ROS production and inflammation when an excess dose of bilirubin nanoparticles is administered.

In this study, exceeding the normal limits of IBIL and DBIL significantly increased the risk of POMC. However, the observed dose-response relationship was solely evident in the unadjusted and adjusted models of unconjugated, lipid-soluble bilirubin (IBIL), potentially due to its lipid-solubility and ability to elicit a robust level of reactive oxygen species (ROS) at low concentrations. Unlike water-soluble direct bilirubin (DBIL), IBIL can permeate any cell by virtue of its ability to readily diffuse through the phospholipid bilayer, thereby influencing mitochondrial reactions, augmenting ROS levels, and even provoking autoreactive immune cell activation.

Our study has several limitations. First, this was a retrospective, single-center study; therefore, different zones and ethnic groups should be considered. Second, although multiple reported POMC risk factors were collected, the reason for the increased IBIL levels in patients with POMC remained unclear. Abnormal liver function and some erythroid cell subsets with immunomodulatory functions should be analyzed during MG progression. Third, the correlation between high IBIL levels, ROS levels, and immune cell activation should be explored in an in vitro experiment, and further in vivo experiments are needed to confirm this mechanism. Finally, this study only concerned MG status during inpatient care, so a longer observation period would be useful to evaluate the association between high IBIL levels and MG progression. The IBIL level in myasthenic crisis that is not caused by thymectomy should be further confirmed.

We found that high preoperative IBIL levels, especially those exceeding the normal range (≥14 μmol/L), are an easily accessible and cost-effective biomarker for predicting POMC incidence. The relationship between IBIL levels and MG progression should be investigated over a longer follow-up period and through in vitro functional experiments.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by the Ethics Committee of Tangdu Hospital, Fourth Military Medical University (reference number: 202111-16). The studies were conducted in accordance with the local legislation and institutional requirements. The Ethics Committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants' legal guardians/next of kin.

CS: Conceptualization, Data curation, Funding acquisition, Investigation, Supervision, Writing—original draft, Writing—review & editing. ZR: Methodology, Supervision, Writing—original draft. YZ: Supervision, Writing—original draft. RG: Data curation, Investigation, Writing—original draft. HL: Data curation, Investigation, Writing—original draft. TW: Data curation, Methodology, Writing—original draft. TG: Data curation, Investigation, Writing—original draft. YT: Data curation, Investigation, Writing—original draft. NS: Formal analysis, Writing—original draft. SH: Data curation, Project administration, Writing—original draft. XH: Investigation, Project administration, Writing—original draft. SL: Data curation, Project administration, Writing—original draft. FN: Data curation, Project administration, Writing—original draft. YS: Data curation, Project administration, Writing—original draft. QL: Supervision, Writing—original draft. QW: Investigation, Project administration, Writing—original draft. XC: Investigation, Project administration, Writing—original draft. ZL: Supervision, Writing—original draft, Writing—review & editing. TC: Project administration, Supervision, Writing—review & editing, Writing—original draft.