Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has triggered a global panic and posed a significant public health threat (1). The phenotype of COVID-19 including susceptibility, hospitalization and severity showed a high degree of heterogeneity (2). Specifically, most patients presented as asymptomatic, and a small proportion required hospitalization or even death. Changes in the neurological and psychiatric sequelae of COVID-19 survivors were tracked (1). The COVID-19 pandemic spans less than four years to date. Therefore, the physical and mental effects of COVID-19 over a longer timeline are still unknown.

Accumulating evidence suggests that SARS-CoV-2 can penetrate the blood-brain barrier and multiple pro-inflammatory cytokines such as IL-6, IL-10, and IFN-γ, trigger a cytokine storm that induces immune dysregulation leading to CNS demyelinating lesions (3–5). These factors are closely related to neuromyelitis optica spectrum disorder (NMOSD). A growing body of case reports or series has documented the emergence of NMOSD subsequent to COVID-19 infection or COVID-19 vaccination (6–9). Although these studies show a clear chronological sequence, it remains uncertain whether COVID-19 infection caused or triggered latent NMOSD. Clarifying the causal relationship between them is urgent in the COVID-19 era.

Mendelian randomization (MR) is a robust approach employed to evaluate the causal relationship between exposure factors and outcomes, leveraging single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) (10, 11). The utilization of the MR approach offers notable advantages in mitigating residual confounding, as the random allocation of genetic variants during conception ensures minimal association with confounding factors. Moreover, this method effectively mitigates the issue of reverse causality, as the genetic variations utilized to approximate the effects of exposure remain unaffected by the occurrence and progression of the outcome (12). Herein, we applied a two-sample MR approach to investigate the causal effects of COVID-19 susceptibility, hospitalization, and severity on NMOSD. The overall design flow of this study is shown in Figure 1 .

In our MR studies, we obtained strongly correlated (P < 5E-8) independent (R² < 0.001) SNPs for exposure. All F-statistics were greater than 10 (from 729 to 99007). These SNPs explained between 1.73% and 24.08% of the variance of exposure. All MR estimates of power were greater than 0.8, representing sufficient ( Supplementary Table S1 ).

To our knowledge, we have comprehensively assessed the causal association between COVID-19 phenotypes and NMOSD for the first time. In this MR study, our findings showed that genetically predicted COVID-19 susceptibility was associated with a high risk of AQP4+ NMOSD. Extensive sensitivity analysis was consistent with the primary results, consolidating the stability and consistency of the MR analysis.

Accumulating evidence shows an inextricable association between SARS-CoV-2 infection and neuroimmune diseases (4). The underlying pathogenesis is that SARS-CoV-2 infection of multiple cells, including macrophages, neutrophils, and dendritic cells, induces an immune response in T cells and B cells by producing interferon and drives the accumulation of pro-inflammatory chemokines and cytokines, creating a “cytokine storm” that invades and damages the central and peripheral nervous system, leading to neuroinflammation and demyelination (4, 25). The development of these processes is related to the mechanisms by which viruses trigger autoimmunity including viral mimicry, propagation of epitopes, activation of bystanders, and presentation of cryptogenic antigens (26).

However, current studies on COVID-19 phenotypes in NMOSD are limited to a few case reports or case series, making it difficult to accurately assess the association between them. Several cases of optic neuritis or transverse myelitis accompanying SARS-CoV-2 infection have been reported (27–30). Other individuals have developed MRI manifestations consistent with NMOSD and positive AQP4 IgG after receiving the COVID-19 vaccine (31–33). More importantly, an increasing number of case reports demonstrate the development of NMOSD in patients after COVID-19 infection, with a proportion detecting positive AQP4 antibodies (8, 9, 34–38). Our study found the causal effect of COVID-19 susceptibility on AQP4+NMOSD, which supports previous observational findings. But the underlying mechanisms need to be clarified. Notably, although we did not find evidence for causal effects of COVID-19 phenotypes on NMOSD or AQP4-NMOSD, we had sufficient power (>0.8) to assess these associations in the current study.

To the best of our knowledge, there are currently no studies utilizing MR approach to investigate the causal effects of other infections on NMOSD. However, a recent review has highlighted the association between COVID-19 and central nervous system demyelination, emphasizing the critical importance of establishing a causal relationship (4). MS, as a distinct entity of demyelinating disorders separate from NMOSD, has received considerable attention. Despite several studies using MR methods to investigate the association between various COVID-19 phenotypes and MS, research results have yielded inconsistent conclusions. For example, Shang et al. investigated the causal effect of COVID-19 using phenotype data from the COVID-19 Host Genetics Initiative (Release 6), but found no positive association with MS (39), in agreement with the conclusions of Larsson et al. (40). However, using the largest available dataset from the COVID-19 Host Genetics Initiative (Release 7), Baranova et al. showed that genetic susceptibility to hospitalized COVID-19 had a causal effect on MS (OR: 1.15, 95%CI: 1.02-1.30, p = 0.022, FDR = 0.044) (41). The use of the most comprehensive COVID-19 phenotype data to date by Baranova et al. makes their findings more reliable and consistent with certain case reports or case series observations (6, 42, 43).

It is important to note that although MR studies have genetically confirmed the potential causal relationship between COVID phenotypes and NMOSD or MS, further validation of these findings is needed, particularly in the absence of large cohort studies. This means that MR studies can be considered as preliminary research that to some extent fills these gaps.

The advantage of the current study is that compared to traditional observational studies, we have utilized MR methods to minimize the interference of reverse causality and residual confounders. Sensitivity analysis also supported the robustness of MR estimates.

Nevertheless, several limitations exist in our study. First, our GWAS data were mainly from European populations, so estimates of the overall population still need to be clarified. Second, we could not specify the proportion of participants with a possible overlap in exposure and outcome. Third, due to the different origins of data on different COVID-19 phenotypes, individuals affected by COVID-19 may manifest phenotypic variation influenced by local factors, such as the standard of health care. Fourth, the genetic explanatory power of significant SNPs is not entirely satisfactory, suggesting the potential involvement of other factors in the occurrence of NMOSD. Due to the large sample size, this study achieved an ideal statistical power (>0.8), which implies the reliability of the statistical conclusions derived from the present research. Fifth, due to data limitations, it becomes challenging to present the clinical characteristics of COVID-19 phenotype data, such as age and gender. This limitation may hinder the clinical applicability. Ongoing MR studies have begun preliminary explorations of the causal effects of different COVID-19 phenotypes on NMOSD, which warrant further verification through cohort studies.

In conclusion, we provide strong evidence supporting the genetic liability to COVID-19 susceptibility was positively associated with the high risk of AQP4+ NMOSD. These contribute to our understanding of the prevention, diagnosis, treatment, and potential pathogenesis of NMOSD during the COVID-19 pandemic. Further studies are still warranted.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The overall research was conceptualized and designed by HZ. DS performed the data analysis and contributed to the writing and revision of the manuscript. All authors participated in data collection and monitoring. The submitted version of the manuscript was reviewed and approved by all authors.