DDX3X, WDR45 and the ongoing mystery of X-linked disorders

X-linked. Almost a decade ago, the former EuroEPINOMICS team was asked to perform a difficult task. We reviewed inherited variants in X-linked genes, trying to understand whether inherited variants are causative of neurodevelopmental disorders for one of our research studies. In most cases, we decided that we did not have enough evidence. How could we tell whether variants in genes such as HUWE1 or CNKSR2 that were transmitted from unaffected females to affected sons were disease-causing or not? I remembered my frustration when I came across a publication on the contribution of X-linked variants to neurodevelopmental disorders that was published last year.

Figure 1. X-chromosomal genes with three or more de novo variants in the DDD cohort, reconstructed from Martin et al., 2021. De novo variants in DDX3X are twice as common as the de novo variants in the next common gene (MECP2). There are only 9 genes with 10 or more de novo variants and 155 X-chromosomal genes have only three de novo variants or less. This makes it difficult to assess whether these genes are truly causative or whether the de novo variants occurred by chance. Martin et al. applied various methods to assess the impact of X-linked coding variants.

DDD. The study by Martin and collaborators examined exome data from 11,044 patient-parent trios from the DDD study that we have discussed in several blog posts before. The authors examined the burden of X-linked coding variants and examined the contribution of these variants to neurodevelopmental disorders. Here are my five take-aways from their study.

1 – We know the most common X-linked genes
The study by Martin used a statistical approach to provide information about genes that were significantly associated with neurodevelopmental disorders. Out of 155 X-linked genes with de novo variants, 23 genes were significant. All of these genes were already established causes of neurodevelopmental disorders.

2 – DDX3X continues to surprise me
I looked back at the blog posts that I wrote exactly a decade ago. WDR45 and MECP2 were already known, and IQSEC2 was just in the process of being discovered. But DDX3X? This gene was only described in 2015 and represents the most common X-linked gene with de novo variants. The number of individuals with de novo variants in DDX3X was the same as the next two conditions (MECP2, WDR45) combined. This gene is a good reminder that there are still a large number genetic etiologies to be characterized that may be more frequent that we would initially expect.

3 – Most X-linked genes are known
While we currently believe that there are still more than 300 genes for neurodevelopmental disorders out there, the study by Martin and collaborators suggests that this is not the case for X-linked disorders. While as yet unknown autosomal genes would explain more than 20% of the missing heritability in neurodevelopmental disorders, this expected proportion for X-linked genes is less than 2%.

 4 – Monogenic causes do not account for the male bias
We would typically assume that the so-called male bias, the greater number of boys with developmental disorders compared to girls, is likely explained by the fact there are X-linked recessive disorders. However, Martin and collaborators find that this is not the case. The identified variants in known X-linked disorder do not account for this bias. In fact, the strict separation of X-linked dominant and X-linked recessive disorders may not be meaningful at all. In both males and females, X-linked disorders account for ~6% of all neurodevelopmental disorders.

5 – Only ~10% of Inherited X-linked variants are causative
Finally, an answer to my question. Only 13% of inherited missense variants in X-linked genes are actually pathogenic. While X-linked de novo variants and inherited protein-truncating variants have a very high positive predictive value (PPV), inherited missense variants do not. When the authors apply strict filters such as CADD scores > 25 and MPC > 2, they are able to increase the PPV to higher levels. However, when the authors compare their results with ClinVar, they estimate that using current diagnostic practices, an estimated 60% of pathogenic missense variants are not classified as pathogenic. In brief, inherited missense variants in X-linked genes continue to be difficult. One option of overcoming these difficulties is to do additional testing in the family, which would allow for excluding a significant number of variants that are found in unaffected males.

What you need to know. X-linked genes account for roughly 6% of all neurodevelopmental disorders equally in males and females and the vast majority of these genes are known. DDX3X, MECP2, WDR45, and IQSEC2 are the most common genes with de novo variants. Inherited missense variants in boys can be reliably classified as pathogenic if these variants are protein-truncating. For inherited missense variants, only 10% are disease-causing.


Ingo Helbig is a child neurologist and epilepsy genetics researcher working at the Children’s Hospital of Philadelphia (CHOP), USA.