Hereditary spastic paraplegias (HSP) are a heterogeneous group of neurodegenerative disorders, characterized by a large phenotypic and genetic heterogeneity [1]. Clinically, two forms are distinguished: uncomplicated and complicated forms. The uncomplicated forms include progressive spasticity and lower limbs (LL) weakness; whereas the complicated forms associate additional neurological signs such as cerebellar ataxia, intellectual disability, peripheral neuropathy, optic atrophy, or brain anomalies on MRI [1]. Moreover, an inter- and intra-familial variability of the phenotype has been reported, regarding age at onset, the presence of associated neurological signs or severity of handicap. To date, more than 70 spastic paraplegia genes (SPG) have been identified [2]. Variants in SPAST (SPG4) are observed in ~40% of HSP with autosomal dominant inheritance. SPG4 is typically uncomplicated, although complicated forms also exist [3]. Incomplete, age-dependent penetrance is observed in families [4]. SPAST variants are frequent in sporadic HSP but the proportion of cases linked to de novo variants is low [5, 6]. Similarly, somatic mosaicism is rare with only three reported cases in three independent series of patients [7,8,9]. We report here the first series of SPAST mosaic variants in four unrelated patients.

Material and methods

Patient’s collection and variants detection and analyses are detailed in Supplementary data.

Clinical and genetic findings

We report in this study four patients with mosaic SPAST variants from four unrelated families (Fig. 1), identified as part of routine diagnosis in the two genetics departments, at Pitié-Salpêtrière and Bordeaux University Hospitals. Three patients have HSP symptoms and one patient is asymptomatic (Table 1).

Fig. 1: Family pedigrees and schematic representation of identified variants in SPAST.
figure 1

A Pedigrees of the four families. Mosaic mosaic variant, homogeneous homogeneous variant, WT wild type; *sampled individual. Black: patients presenting spasticity. B schematic view of SPAST with its main domains and the variants found in the series.

Table 1 Clinical and paraclinical examinations and genetic findings in the different patients.

In Family 1, the proband (1.II.2) is a 9-year-old boy presenting a moderate spasticity since the age of 1 year, associated with moderate motor deficit and mild amyotrophy of the LL. Next generation sequencing identified in patient 1.II.2 the previously reported NM_014946.3:c.1496G>A p.(Arg499His) recurrent variant in exon 13 [4, 5, 10, 11]. Allelic frequency (AF) was 26% (143/543 reads) on first NGS identification and 32% on an independent sequencing (194/604 reads). Estimation of mosaicism rate was hence around 58% of patient’s lymphocytes and statistically significant (chi-squared test, p value < 0.00001). This variant is located on the AAA domain of the spastin protein, where missense variants have previously been associated with an earlier age at onset [2]. Segregation studies could only be performed in the mother (1.I.2) who did not carry the variant. This variant was previously reported in at least 4 symptomatic individuals. Two of them presented an early pure and isolated form of SPG4 as observed in our patient [5, 11]. The third patient presented a pure familial form and the last one had an early complex form with dysarthria leading to anarthria at age 12 years old [4, 10].

In Family 2, the proband (2.II.1) presented a mild spasticity with late onset (at 60 years) associated with moderate motor deficit and mild amyotrophy at LL. Brain MRI revealed slight periventrical white matter hyperintensities. NGS identified a novel variant, NG_008730.1(NM_014946.3):c.1413 + 1_1413 + 2del, which was predicted to affect splice donor site of intron 11 (predicting scores: SpliceSiteFinder = 75.37, MaxEntScan = 8.07 and NNSPLICE = 0.97). Observed read fraction was 19% (164/847 reads) on first NGS identification and 17% (229/1329 reads) on an independent sequencing. Mosaicism was evaluated at 36% of patient’s lymphocytes (p value < 0.00001). Parental DNAs were unavailable for genetic testing.

The index case of Family 3 (3.II.1) presented since childhood a complex phenotype associating moderate spasticity, weakness of the LL, mild LL amyotrophy, mild ataxic gait, and infantile hypertrophic pyloric stenosis. An array-CGH was performed to investigate the uncommon association between HSP and hypertrophic pyloric stenosis. Two microdeletions were identified at the heterozygous state: a 2p22.3 613 kb microdeletion NC_000002.11:g.(32289091_32312562)_ (32958959_32999928)del, disrupted SPAST from exons 2–17, and a 512 kb 3q25.2q25.31 microdeletion including MME. MME (OMIM*120520) had been implicated in autosomal‐recessive Charcot–Marie–Tooth disease [12]. This gene has also been associated with the autosomal dominant spinocerebellar ataxia 43, but in a single large Belgian family [13]. Analyses were performed in the two parents and showed that the deletions were both inherited from the father. Notably, SPAST deletion was observed in the healthy father with a mosaic pattern on array-CGH (log2ratio = −0.31, Fig. 2) whereas the 3q25.2q25.31 deletion was observed with homogeneous heterozygous pattern (log2ratio = −1). A karyotype excluded a balanced translocation. Mosaicism of the SPAST deletion was confirmed by interphase and metaphase FISH analysis and evaluated to 45% of father’s lymphocytes and AF estimated to 23% (p value < 0.00001). At the age of 68 years, the father was not complaining of any symptoms and had a normal electromyogram.

Fig. 2: Array-CGH profiles of patient 3.II.1, and his father 3.I.1. Probes are plotted as dots corresponding to the log2 ratio of fluorescence intensities Cy5/Cy3.
figure 2

Observe the deletion in the proband 3.II.1, partially including SPAST (log2 ratio is equal to −0.948). In the father’s proband (3.I.1), log2 ratio is around −0.31, indicating a deletion with a mosaic pattern (mosaic rate estimated around 45% and allelic frequency around 23%).

The index patient (4.II.2) of Family 4 presented with a severe spasticity of very late onset (at 70 years) associated with moderate LL amyotrophy. His daughter (4.III.2) had a milder phenotype with an earlier onset at 47 years old. NGS analyses were performed in two individuals and identified the previously reported variant, NM_014946.3: c.67_85dup p.(Leu29Glnfs*25) in exon 1 [14, 15]. Index case (4.II.2) carried the variant with a mosaic pattern. Mosaicism was evaluated at 58% of all father’s lymphocytes, with an AF of 30% (29/98 reads, p value < 0.01) in first NGS identification and 28% in an independent sequencing (103/371 reads). As expected, the variant was observed at homogeneous heterozygous state in the daughter. This variant was previously reported in 2 patients, one with a pure familial form, and one with a pure sporadic form [14, 15].


We report four unrelated patients with a SPAST variant at the mosaic state. Mosaic SPAST variants have been rarely reported in the literature, with only three patients described so far [7,8,9]. Our study confirms this low frequency, with four patients identified in more than 300 patients with a SPAST variant previously reported by our diagnostic laboratories [2]. However, a large segregation study in each family was not achievable (most often due to the absence of parental samples, sometimes deceased, or not wishing to have samples taken) and we cannot exclude the existence of other mosaic carriers among the patients’ relatives in our cohorts.

SPAST variants are the most common cause of autosomal dominant pure HSP, and only 5% of SPAST mutated cases are sporadic [2]. In these sporadic cases, confirmation of the de novo status of the variant is rarely established. In a large review of the literature in 2019, Schieving et al. found only 14 patients with de novo variants [6]. This frequency appears low compared to the large number of publications describing individuals, symptomatic or not, carrying SPAST variants. We hypothesized that the very low frequency of SPAST mosaic cases could be explained by a non-systematic exploration of parents, especially for patients with late onset.

Based on data from the literature and our series, we note that there are more men with mosaicism than women, with only one women described with a SPAST mosaicism, who was asymptomatic [8]. It is established that SPG4 has a lower penetrance in women, the hypothesis being that in women low allelic fractions may not reach the threshold for phenotypic expression as easily as males can do [2].

Three patients with SPAST mosaic variants were affected whereas one patient was asymptomatic. All symptomatic mosaic carriers had an uncomplicated HSP form. This observation conforms with previously reported data where complicated HSPs are rare in SPAST variants carriers [2, 16]. Several studies have reported asymptomatic SPAST variant carriers, representing ~6% of all molecularly confirmed patients [2, 4, 5, 17]. Asymptomatic mosaic carriers have been previously reported but with a very low rate of somatic mosaicism in lymphocytes [8]. In our study, all patients with mosaic variant presented a high rate of mosaicism, from 36 to 58% of lymphocytes. Moreover, asymptomatic patient 3.I.1 did not present the lowest rate of mosaicism. We could not determine a minimum rate of mosaicism of SPAST variant triggering the disease. However, blood mosaicism may not reflect mosaicism rate in other tissues, especially in the nervous system [18]. Since this patient was 68 years old at examination, a later onset of HSP for this patient cannot be excluded, as previously reported for this type of truncating variants [2]. Finally, the boundary between a mosaicism and a germ-line heterozygosity can be tenuous, with very close allelic frequencies. This notion must be taken into account in the interpretation of allelic frequencies close to 50%.

The nature of the variant does not seem to be associated with mosaicism since three out of four mosaic pathogenic variants reported here are variants leading to premature stop codons including three frameshift variants and one large deletion, and previously reported mosaic patients included missense and nonsense variants, showing that all variant types that affect function can be found at mosaic state [7,8,9].

In conclusion, patients with mosaic SPAST variants appear to be rare. Nevertheless, systematic screening of both parents of apparently sporadic patients and assessment of the allelic imbalance could increase detection of mosaicism and clarify the number of individuals carrying mosaic variants, which is of great importance for genetic counseling. Future research may also be focused on understanding the role of sex in modulating the mosaicism expressivity [19].