Cases involving partial deletions or duplications of chromosome 18 are well documented in the literature. The 18q- syndrome constitutes one of the frequent autosomal deletion syndromes in man, with more than 100 patients reported [1]. The syndrome includes moderate intrauterine growth retardation, moderate mental retardation, and a specific pattern of dysmorphisms and anomalies [1]. Mosaicism for a deleted chromosome 18 has been described in a few patients with mostly the full clinical picture of the 18q- syndrome. Here, we report a patient with an unusual mosaic karyotype consisting of cells with normal karyotype and others with a terminal deletion of one chromosome 18 and the other chromosome 18 having an interstitial duplication.

The 18q- syndrome is one of the more frequent autosomal deletion syndromes in man, with more than 100 patients reported [1]. This deletion involves del(18)(q21-qter) with the majority of breakpoints in 18q21.2-q22.2 [4]. About 75% of patients have de novo deletions with the remainder having unbalanced translocations or inversions. Of the de novo cases, 85% are paternal in origin [5]. The degree of mental retardation, growth retardation and malformations does apparently not correlate with the breakpoint position [1]. Most patients with subband determination have the breakpoint at 18q21.3 and these patients rarely have congenital malformations.

The syndrome comprises moderate intrauterine growth retardation, facial dysmorphism, moderate mental retardation, and various anomalies [1]. The facial dysmorphism may include brachycephaly, midface hypoplasia, everted lower lip, narrow upper lip with absent philtrum, downturned corners of the mouth, depressed and wide nasal bridge, small teeth, deep-set eyes with downslanting palpebral fissures, epicanthus, strabismus, prominent anthelix and antitragus, and very narrow external ear canal [1]. The associated anomalies include kyphosis/scoliosis, vertebral anomalies, genital hypoplasia, proximally placed thumbs, tapering fingers, clinodactyly of fifth fingers, coxa valga, clubfeet, toe anomalies, dimples over joints, transverse palmar creases, congenital heart defects, hypospadias, cleft lip and palate, deafness, and several less frequent anomalies [6, 1]. A recent study using high-resolution array-CGH in 29 patients with such deletions concluded that the critical region for the 'typical' 18q- phenotype is a region of 4.3 Mb located within 18q22.3-q23 [7].

The neuropsychiatric phenotype has been characterized in 27 patients with age range 2–47 years [8]. Intelligence ranged from borderline to severely deficient. Performance in specific neuropsychological functions was consistently in the moderately-to-severely impaired range. Behavioral problems were common and 6% of the patients had autism. There was no relationship between deletion size and any measure of cognition or behavior [8].

Mosaicism for a deleted chromosome 18 was repeatedly found in patients with mostly the full, and occasionally a diminished clinical picture of the 18q- syndrome. Partial monosomy and partial trisomy involving different segments of chromosome 18 in the same patient have been described in pericentric inversion recombinants [9]. The patient of an early report had a clinical diagnosis of 18q- syndrome and a recombinant chromosome 18 with deficiency of the distal one third of 18q and duplication of the terminal segment of 18p [9]. The mechanism here involves crossover within the formed inversion loop, so called 'aneusomie de recombinaison' [10]. A maternal paracentric 18q inversion gave rise to a recombinant del/dup18q in the twin offspring [11]. Molecular cytogenetic and molecular analyses suggested that the duplication-deletion chromosome resulted from breakage of a dicentric recombinant chromosome 18 [11]. The same mechanism of an intermediate dicentric chromosome undergoing breakage and giving rise to a deleted and an inv dup chromosome, has been demonstrated for mono- and dicentric 8p duplications [12].

A mosaic karyotype 46,XY,del(18)(q21.3q22.2)/47,XY,del(18)(q21.3q22.2)+r(18) was found in a mentally retarded male with a mild form of the 18q- syndrome [13].

A mentally retarded boy with 46,XY,del(18)(q22)/46,XY,psu idic(18)(q23) had clinical manifestations compatible with the 18q- syndrome [14]. A possible mechanism involving an early postzygotic U-type exchange of sister chromatids at 18q23 (yielding an unstable dicentric chromosome), breakage at 18q22 (resulting in the 18q- chromosome of one cell line), and inactivation of one centromere (giving a stable pseudoisodicentric chromosome 18 in the other cell line) was presented [14]. A mosaic isopseudodicentric chromosome 18q was detected in a girl with discrete features of trisomy 18 [15]. The 18q23 breakpoint was very distal as FISH with one of the most telomere located probes could not confirm monosomy of distal 18q23 [15]. A few cases of isopseudodicentric chromosome 18q have been reported and most of the cases are mosaic, although one non-mosaic case has been described recently [16]. In that case, the isopseudodicentric chromosome 18q was present in 100% of blood lymphocytes, and a terminal deletion 18q with identical 18q22.1 breakpoint as the i(18q), was detected in placental cell cultures [16]. The authors pointed out that the child represented the first non-mosaic case being viable, although with severe congenital malformations and neonatal death at 14 weeks of age [16].

Three cell line mosaicism including full trisomy 18, normal cells and the majority of cells with partial trisomy involving an extra chromosome 18 deleted at band q22, was described in a girl with cardiac and CNS anomalies, dysmorphic facial features, failure to thrive, and developmental delay [17]. Non-disjunction studies of trisomy 18 have demonstrated that an error in maternal meiosis II (MII) is the most frequent cause of non-disjunction leading to trisomy 18 [18], unlike human trisomies 13, 15, 16, and 21, which show a higher frequency of maternal meiosis I (MI) errors [19]. In the present case, microsatellite analysis indicated maternal meiotic origin of the extra chromosome 18 and paternal origin of the deleted chromosome 18 [17]. The authors suggested a trisomic conception followed by two early postzygotic (mitotic) errors as the simplest mode to explain the karyotype of this patient.

A baby girl with severe growth retardation, developmental delay, microcephaly, and multiple congenital anomalies, had a mosaic karyotype consisting of cells with an i(18q) and another cell line with a r(18q) [20]. The authors noticed that the isochromosome was probably formed by centromere misdivision with loss of the i(18p) chromosome. They further stated that the rearrangement had arisen before conception and suggested that the ring chromosome originated from the cells with an i(18q) chromosome by breakage in both long arms with subsequent reunion, a mechanism which had been previously described in ring chromosome 21 formation [21]. A recent study on ring chromosome formation using array-CGH showed not only the expected terminal deletion but also a contiguous duplication through a classical inv dup del rearrangement [22].

In Fig. 3 we present a schematic drawing of the comprehensive results of GTG banding, FISH and array-CGH in the proband. The FISH results are in agreement with the cytogenetic findings, if the 18q21.2 duplication represented by BAC clone RP11-160B24 is small. The 18q deletion detected by array-CGH is in agreement with the size of the cytogenetic deletion, if we assume that 18q21.3-18qter is deleted on the one chromosome. The finding by array-CGH of a relatively small not-deleted region (about 2.2 Mb) within the deleted region is intriguing, and can be interpreted as a duplication located on the other chromosome. A 2.2 Mb duplication would not be visible by ordinary GTG banding, neither by FISH if probes are not used within the duplicated area. The array-CGH cannot discern whether the 2.2 Mb duplication is located on the chromosome with the terminal deletion or on the other chromosome with the duplication represented by BAC clone RP-160B24. The extent of the FISH duplication was not further investigated and was not picked up by array-CGH.

Low copy repeats (LCRs) spread over the human genome are stretches of duplicated DNA more than 1 kb in size. Non-allelic homologous recombination between highly similar LCRs has been implicated in numerous genomic disorders [23], and LCRs have also been characterized at breakpoints of both balanced and unbalanced chromosomal rearrangements [24]. It is interesting to note that in our patient, BAC clone RP11-160B24 showed a duplication in band 18q21.2, with the deletion breakpoint being located in adjacent band 18q21.31 (as determined by array-CGH). The use of the array-CGH technique has revealed the unbalanced or complex nature of a proportion of de novo apparently balanced translocations [25]. In 27 cases of de novo reciprocal translocations in patients with an abnormal phenotype, 11 (40%) were found to be unbalanced after array-CGH analysis [25]. Furthermore, 5 out of 27 (18%) of the reciprocal translocations were instead complex rearrangements with more than 3 breakpoints [25].

It is important to keep in mind that, like in our case, an unexpected duplication in deletion cases might confuse the genotype-phenotype correlation, and array-CGH continues to be an important technique in characterizing chromosomal rearrangements, their mechanisms of formation, precise establishment of genotype-phenotype correlations, and genetic counseling.

Written informed consent was obtained from the parents of the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.