Do not jump to easy conclusions! Lessons from pitfall in the molecular diagnosis of ARSACS

To the Editor : We present our recent experience concerning an initial pitfall in the molecular characterization of two distant relatives affected by Autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) (1, 2) (Fig. 1), and corrected after extending the analyses to informative family members. In particular, we underscore the need for caution in interpreting molecular findings when segregation cannot be easily determined; this issue is critically related to the emerging occurrence of variants with unclear significance while sequencing SACS gene in patients with spastic ataxias (1, 2), as well as to the wider use of whole-exome sequencing (WES) in clinical settings, which also increases the risk of possible drawbacks in molecular diagnosis. Case 1 (Fig. 1) has been fully reported elsewhere (3). Briefly, a 26-year-old man suffered from a progressive spastic ataxia, cerebellar dysarthria and dysmetria since his teens. Electromyography (EMG) studies documented a mixed sensory-motor polyneuropathy, and a brain magnetic resonance imaging (MRI) showed features typical of ARSACS (3). Direct sequencing of SACS revealed two likely loss-of-function mutations (c.4108C>T/p.Q1370* and c.6837insA/p.K2279fs11*; Fig. 1). Both variants predicted prematurely truncated proteins with a likely deleterious effect in silico [upon scrutiny of several prediction softwares including, among others, mutation taster (www.mutationtaster.org/) and polyphen2 (genetics.bwh.harvard.edu/pph2/)]. They were also not listed in large polymorphic databases (dbSNP, www.ncbi.nlm.nih.gov/projects/SNP; Exome Variant, evs.gs.washington.edu/EVS/). At that time, we all were pleased with the clinical and molecular diagnosis, and assumed that the two variants were associated with the disease. However, the following evaluation of case 2, another family proband, showed that our conclusion had been misled or, at least, too premature. Case 2 (Fig. 1) was recently examined by us at age 34. At that time, she was unaware of case 1 clinical status, and her kinship with him was not revealed during the preliminary interview. Case 2 had a normal psychomotor development; scissoring gait was initially noted around age 10 and had slowly worsened over time. She had done well at school, and has been working as an administrator for a business firm. Her neurological examination, EMG and MRI features, were consistent with a diagnosis of ARSACS (2). Upon collection of a thorough family history, her distant relationship with case 1 prompted us to seek right away for the same SACS changes detected in case 1. However, only the c.4108C>T/p.Q1370* change was discovered, prompting full analysis of SACS and identification of two further changes (Fig. 1). The newly discovered variants in SACS (namely, c.11012_11013delAA/p.Q3671Rfs23* and c.11598delC/p.G3866fs3*) were absent in large variant databases and predicted to be highly deleterious in silico. To clarify the segregation pattern of the mutations found in this family, having excluded somatic mosaicism for the specific mutations, we fully reanalyzed the SACS gene in case 1 using different primers. In addition to the c.4108C>T/p.Q1370* and c.6837insA/p.K2279fs11* already recorded at the time of our premature report (3), analyses also disclosed the c.11598delC/p.G3866fs3* possibly missed in earlier studies because of allele drop out. Therefore, we sought out the three variants in the probands’ living healthy parents: the mother from case 1 harbored the heterozygous c.6837insA /p.K2279fs11* whereas the father showed the heterozygous c.4108C>T/p.Q1370* and c.11598delC/p.G3866fs3*, likely in cis-configuration (Fig. 1). On the basis of these findings, we could definitively correct the genotype of the two patients (Fig. 1), though we are unable to say which of the two changes [c.4108C>T+c.11598delC] on the paternal allele was more deleterious to protein function or if their cisconfiguration further affected sacsin. This report shows some key points. The assumption of identical genotypes in the presence of highly similar phenotypes might be misleading even within a single family, emphasizing the need to perform full gene analysis in patients, while extending molecular analyses to their parents in order to test allelism of potential gene variants. In particular, this must be done to exclude cisconfigurations of apparently compound heterozygous variants, as in our case.