Detection of base substitution-type somatic mosaicism of the NLRP3 gene with >99.9% statistical confidence by massively parallel sequencing

Chronic infantile neurological cutaneous and articular syndrome (CINCA), also known as neonatal-onset multisystem inflammatory disease (NOMID), is a dominantly inherited systemic autoinflammatory disease and is caused by a heterozygous germline gain-of-function mutation in the NLRP3 gene. We recently found a high incidence of NLRP3 somatic mosaicism in apparently mutation-negative CINCA/NOMID patients using subcloning and subsequent capillary DNA sequencing. It is important to rapidly diagnose somatic NLRP3 mosaicism to ensure proper treatment. However, this approach requires large investments of time, cost, and labour that prevent routine genetic diagnosis of low-level somatic NLRP3 mosaicism. We developed a routine pipeline to detect even a low-level allele of NLRP3 with statistical significance using massively parallel DNA sequencing. To address the critical concern of discriminating a low-level allele from sequencing errors, we first constructed error rate maps of 14 polymerase chain reaction products covering the entire coding NLRP3 exons on a Roche 454 GS-FLX sequencer from 50 control samples without mosaicism. Based on these results, we formulated a statistical confidence value for each sequence variation in each strand to discriminate sequencing errors from real genetic variation even in a low-level allele, and thereby detected base substitutions at an allele frequency as low as 1% with 99.9% or higher confidence.