In the present study we have characterized on the genetic aspects and clinical ground the largest cohort of ring 20 patients reported till now. All patients but the youngest one (MM, 3 years old) presented the typical seizure disorder associated with r(20) chromosome, with a mean age at seizure onset of 9 years. Epilepsy onset has been reported in paediatric, adolescent and adult age. Patients with an early epilepsy onset (within the first year of age) have been reviewed by Ville et al: all exhibited severe psychomotor delay before seizure onset and a percentage of r(20) chromosome cells ranging from 87 to 100% .
More than 60 cases of patients with a non supernumerary r(20) chromosome mosaicism have been reported, but the great majority have been investigated only with conventional cytogenetic analysis.
Ring(20) syndrome is characterized by phenotypic variability, particularly concerning dysmorphism, malformations, mental retardation and behavioural disturbances . The severity of clinical features has been ascribed to high percentages of r(20) chromosome abnormal cells, ring instability and presence or absence of chromosome 20 deleted regions [10, 13, 21]. Previously reported patients showed a percentage of r(20) chromosome cells ranging from 1 to 100% in peripheral blood. Generally, cases with 100% anomalous cells exhibited more severe development and psychomotor delay, starting from the first months of life, before seizure onset. Previous studies failed to reach a consensus regarding the relationship between the percentage of mosaicism and the manifested phenotype. No correlation was found between IQ and age at seizure onset in two independent reports,[6, 9] whereas Nishiwaki et al.  concluded that the mosaicism ratio is significantly associated with IQ, age at seizure onset and malformations, but not with the response to antiepileptic drug treatment. A positive correlation between age at seizure onset and rate of r(20) mosaicism was also observed by Herrgard et al. . In our patient cohort the percentage of r(20) chromosome cells covered a wide range, from 8 to 71% in peripheral blood. In seven out of eight investigated cases the percentage of r(20) chromosome could be also computed on fibroblasts, making the level of mosaicism representative of two different cell lineages. Our results indicate that higher percentages of r(20) chromosome cells are observed in probands with younger age at seizure onset and seem also to correlate with resistance to antiepileptic drug treatment, agreeing with the findings of Nishiwaki et al. and Herrgard et al. [10, 11]. The two patients from our series showing a lower degree of r(20) mosaicism developed seizures at an older age, epilepsy was controlled and they are the only two patients without mental impairment in adult age. These cases further demonstrate the fundamental role of seizures in determining cognitive impairment in r(20) syndrome . Behavioural problems also appear in our patients to be related with higher percentages of r(20) chromosome cells. A major malformation was observed only in BV, who is the patient with a concurrent 1.9 Mb deletion on 16q chromosome (Figure 4a and 4b), confirming that this finding is likely to be coincidental and not characteristic of the r(20) mosaic patients in addition to the dysmorphism, which were observed just in BV and MM.
It has been suggested that the presence of chromosome 20 monosomic cells generated by r(20) chromosome instability during mitosis might be implicated in the phenotypic manifestations of r(20) syndrome  as well as in the so call "ring syndrome" . However the FISH results on our series revealed only about 1% of monosomic cells in each patient, different to the significantly higher percentage (8%) detected by Elghezal et al.  but in agreement with the recently reported conclusions weakening the hypothesis of the "ring syndrome" phenotype caused by ring instability . Ring (20) chromosome mosaicism generally represents a de novo event. To date, only three families with several carrier members have been published. A clinically unaffected carrier mother transmitting the r(20) chromosome to two siblings has been reported by Back et al. , while in the familial cases described by Herrgard et al.  the two siblings inherited the r(20) chromosome from their affected mother. Canevini et al. reported a son (FM in this study) inheriting the r(20) chromosome from his less severely affected mother (FMA in this study) . The concomitance of a cell line with two apparently "normal" chromosomes 20 in patients with an inherited r(20) chromosome is still an open question. A possible explanation is that the "normal" cell line is isodisomic for chromosome 20. This condition may have occurred by a rescue mechanism rectifying through duplication of the normal chromosome 20 the monosomic cell originated by loss of r(20) chromosome. Another mechanism leading to a mosaicism with a normal cell line in r(20) chromosome inherited cases is the trisomy rescue in a 47,+r(20) conceptus consequent to a random loss of r(20) chromosome or normal chromosome 20 in different cells during mitosis in early development. We have thus searched for putative UPD20 in our mosaic familial and de novo cases by segregation analysis of chromosome 20 microsatellites from parents to probands. To the best of our knowledge, this is the first study in which UPD20 has been evaluated in patients affected by r(20) disease. Our results do not support the UPD20 hypothesis, although they suggest through the haplotype signature that the r(20) chromosome is the same chromosome that appears and is not circularized in the "normal" cell line (Figure 5). This hypothesis is not contradicted by the FISH results obtained using a 20-specific alphoid probe able to visualize different signal intensities when the two centromeric constrictions are built up with a different number of alphoid repeats (Figure 1). This finding might have the following implications: a) r(20) patients have two cell lines differing only on a morphological but not a genetic point of view; b) one of the two chromosomes 20 has a genetic motif, yet unknown, inducing both the "circularization" and the "re-opening". The direct physical link between distal 20p and 20q segments in the ring structure may disturb the correct expression or the regulation of genes located nearly the telomeric regions. It may also prime the phenotypic manifestation of the syndrome. The mosaic condition of the patient inheriting the r(20) chromosome from his mother may evidence the plasticity of the circularized chromosome.
To date only 10 r(20) patients, including 3 familial cases , have been investigated by FISH techniques [4, 11–13, 21, 25] and one with telomere PRINS (primer in situ DNA synthesis) . Three out of 7 cases were found to carry a r(20) chromosome with deletion of the telomeric regions [13, 21, 26]. In 9 probands, the subtelomeric 20p and 20q regions have been studied and the r(20) chromosome has been shown to maintain these sequences in all carriers [4, 11, 13, 14] but one .
Ring(20) chromosome without deletion of the CHRNA4 and KCNQ2 epilepsy genes was found in the only two patients so far investigated for this aspect [12, 13]. In our series of patients FISH analyses indicated no deletions of the telomeric and subtelomeric chromosome 20 regions or deletion of CHRNA4 and KCNQ2 genes. However, small deletions in their promoters or in regulatory regions can't be excluded. No evidence for loss of chromosome 20 genetic material was also provided by the array-CGH analysis, even if the possibility of a subtle submicroscopic deletion on r(20) chromosome can't be ruled out, since array-CGH may fail to detected low level r(20) cell mosaicism.
Our wide genetic study includes UPD20 evaluation and array-CGH analysis which had never been performed before on r(20) patients. We speculate that the mechanism/mechanisms underlying the seizure disorder in r(20) carriers must be related on the conversion of ring 20 to normal chromosome 20 morphology and vice versa. This speculation should be corroborated by studying further familial cases. Moreover, the hypothesis that tandem repeats of telomere DNA forming a heterochromatic structure may silence subtelomeric CHRNA4 and KCNQ2 genes or their regulatory components should be investigated. The presence of other epileptic genes mapping to chromosome 20 also deserves to be considered in the future within the panel of players in the pathogenetic cascade responsible for r(20) syndrome manifestations.