Background
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by repetitive, stereotypical behaviours and impaired expressive communication [
1]. The prevalence of ASD is estimated to be ~ 2% in developed countries [
2] and relatively lower in lower-income areas [
3]. The aetiology of ASD remains elusive; however, many risk factors for ASD have been discovered, including genetic and environmental factors [
4].
Cytomegalovirus (CMV) is considered a risk factor for ASD since CMV-infected infants can manifest multiple central nervous system (CNS) disorders, such as microcephaly, periventricular calcifications, and sensorineural hearing loss [
5]. As a widely known virus affecting the CNS in paediatric patients, CMV can be transmitted from pregnant mothers to newborns, which is defended as congenital CMV infection. Postnatal infections are usually transmitted by exposure to human milk, blood products, or transplanted organs. Most congenital infections and human milk-associated CMV infections are asymptomatic [
5,
6].
CMV infection has been suggested as a risk factor for ASD for many decades, but systematic studies on the correlation between CMV infections and ASD are still limited. There are some case series that reported that congenital CMV increased the risk of ASD; however, the sample sizes of these studies are very small (n < 10) [
7,
8]. A meta-analysis showed a high prevalence of congenital CMV infection in ASD cases, but this result was still limited by having too few events (0–2) in all included studies [
9]. The titer and seropositivity rates of antibodies to CMV are similar between children with ASD and healthy controls [
10]. Moreover, a 12-year retrospective analysis revealed that cognitive and motor development were similar in extremely premature infants with or without postnatal CMV [
11]. These studies showed that the relationship between CMV infection and ASD remains unclear and needs to be further clarified.
Mendelian randomization (MR) is a method that uses variants that are robustly associated with exposure factors to generate more reliable evidence regarding which interventions should influence specific outcomes [
12]. Evidence derived from MR has a lower strength than randomized controlled clinical trials (RCTs) but is stronger than observational studies, such as case‒control studies [
13]. On the other hand, MR studies usually include a much larger sample size than clinical trials, especially when there is a low clinical prevalence.
In this study, we used 2-sample MR to investigate the association between CMV infection and ASD. We used genetic variants associated with unspecified cytomegaloviral diseases, anti-CMV IgG levels, and maternal CMV as instrumental variables to evaluate CMV infection as a risk factor for ASD.
Discussion
ASD affects people worldwide and brings both health and economic burdens. As a widely known virus affecting the CNS in paediatric patients, CMV has long been suggested as a risk factor for ASD, yet no definite evidence has been found. To investigate whether CMV infection has a causal relationship with ASD, we conducted 3 MR analyses between ASD and unspecified cytomegaloviral diseases, anti-CMV IgG levels, and maternal CMV using GWAS summary data from the FinnGen and IEU Open GWAS projects.
A total of 15 variants were used as IVs in these 2-sample MR studies. Limited by low clinical prevalence, no variants significantly associated with CMV infection (default cut-off used in MR was p < 5 × 10− 8) were identified in these GWAS datasets. This normally used threshold of 5 × 10− 8 would predict an F statistic of approximately 30, and a higher F statistic means lower bias. However, all IVs but 1 (rs146990284) had an F over 10, indicating that they were not too weak instruments.
MR Egger regression, IVW, weighted median, weighted mode and simple mode were used in MR analysis. The results showed no causal relationship between CMV infection and ASD, indicating that CMV IgG positivity could not be considered a risk factor for ASD. Mendelian randomization studies could derive evidence with stronger strength than observational studies with a much larger sample size [
13],thus partly covering the shortage of previously published studies with too few events (0–2) included.
While qualitative studies were limited by event number, quantitative studies were published. Slawinski et al. reported that maternal CMV seropositivity was related to autism symptoms (n = 82 and 68 for CMV-negative and CMV-positive groups,respectively) [
20]. They used the Social Responsiveness Scale, Second Edition (SRS-2) as the ASD evaluation method, and in the maternal CMV sero-positive group, the mean score was 54.8, which was higher than that of the negative group (51.1). However, the population mean of the SRS-2 was expected to be 50, with a standard deviation of 10, and in ASD diagnosis, it was used as a categorical outcome (< 60 indicated normal and 60–65 indicated mild deficits in social interactions). This study indicates that CMV infection might contribute to a small part of ASD risk factors, which is not enough for a categorical outcome.
The main limitation of this study is the small number of IVs, and most IVs are not strong instruments, which is a consequence of the low clinical prevalence of CMV infection. A large number of IVs with strong evidence strength is not always available, and a small number of IVs with strong and specific biological links could also provide evidence. To further assess whether our IVs are informative or not sufficiently informative, we conducted a positive-control MR analysis between CMV and hydrocephalus [
19] (Additional file
1). A significant causal relationship was found between unspecified cytomegaloviral diseases (5 IVs) and hydrocephalus. All 5 IVs from unspecified cytomegaloviral disease GWAS summary data had an F-statistic of approximately 20. However, the maternal CMV dataset with fewer IVs (3 IVs) did not have a positive result in the positive control. The negative results might be a consequence of a low proportion of symptomatic congenital CMV. Approximately 80–90% of CMV infections are asymptomatic [
5,
6]. If only severe infections contribute to the outcome, then the number of events would be very low; consequently, a very large sample size is needed to assess the risk. These factors showed another limitation in this study. Asymptomatic CMV infection and infection with severe symptoms were not distinguished in the exposure datasets. Study 1 with an unspecified cytomegaloviral disease (symptomatic CMV infection) dataset could partly cover this shortage. The combination of all 3 studies could increase the sensitivity of finding causal relationships between CMV infection and other clinical symptoms.
Our results provide evidence that CMV infection does not directly increase ASD risk by 2-sample MR, and this evidence is stronger than observational studies. These results, together with published literature, show the complexity of ASD aetiology. Researchers should be more cautious when analysing the association between CMV infection and ASD. To further investigate the contribution of CMV infection to ASD aetiology, population-based studies with large sample sizes are needed, as is a better-designed attribution analysis.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.