|Home | About | Journals | Submit | Contact Us | Français|
Subtelomeric imbalances are a significant cause of congenital disorders. Screening for these abnormalities has traditionally utilized GTG-banding analysis, fluorescence in situ hybridization (FISH) assays, and multiplex ligation-dependent probe amplification. Microarray-based comparative genomic hybridization (array-CGH) is a relatively new technology that can identify microscopic and submicroscopic chromosomal imbalances. It has been proposed that an array with extended coverage at subtelomeric regions could characterize subtelomeric aberrations more efficiently in a single experiment. The targeted arrays for chromosome microarray analysis (CMA), developed by Baylor College of Medicine, have on average 12 BAC/PAC clones covering 10 Mb of each of the 41 subtelomeric regions. We screened 5,380 consecutive clinical patients using CMA. The most common reasons for referral included developmental delay (DD), and/or mental retardation (MR), dysmorphic features (DF), multiple congenital anomalies (MCA), seizure disorders (SD), and autistic, or other behavioral abnormalities. We found pathogenic rearrangements at subtelomeric regions in 236 patients (4.4%). Among these patients, 103 had a deletion, 58 had a duplication, 44 had an unbalanced translocation, and 31 had a complex rearrangement. The detection rates varied among patients with a normal karyotype analysis (2.98%), with an abnormal karyotype analysis (43.4%), and with an unavailable or no karyotype analysis (3.16%). Six patients out of 278 with a prior normal subtelomere-FISH analysis showed an abnormality including an interstitial deletion, two terminal deletions, two interstitial duplications, and a terminal duplication. In conclusion, genomic imbalances at subtelomeric regions contribute significantly to congenital disorders. Targeted array-CGH with extended coverage (up to 10 Mb) of subtelomeric regions will enhance the detection of subtelomeric imbalances, especially for submicroscopic imbalances.
Abnormalities in genomic copy number are frequently found in patients with multiple congenital anomalies (MCA), dysmorphism, developmental delay/mental retardation (DD/MR), and other neurological disorders [Lee and Lupski, 2006; Shaffer et al., 2006; Aradhya et al., 2007; Baris et al., 2007; Engels et al., 2007; Kriek et al., 2007; Lu et al., 2007; Stankiewicz and Beaudet, 2007]. Rearrangements of chromosome subtelomeric regions represent a high proportion of cytogenetic abnormalities. The largest study of subtelomeric abnormalities to date examined 11,688 unselected cases using subtelomere fluorescence in situ hybridization (FISH) analysis and found an overall abnormality rate of 3.0% with 2.6% deemed pathogenic [Ravnan et al., 2006]. In a recent report of ~7,000 unselected cases using targeted microarray analysis with evolving arrays of varied subtelomere coverage, 2.4% of the cases were determined to have clinically significant subtelomeric rearrangements [Ballif et al., 2007]. Earlier studies that focused on patients with DD/MR indicated an average overall abnormality rate of ~5% and a rate of 2.6% for cryptic (i.e., not resolved by G-banding) subtelomeric aneusomies [Biesecker, 2002; Xu and Chen, 2003; Moog et al., 2005; Yu et al., 2005; Menten et al., 2006].
Substantial phenotypic abnormalities are likely to occur from subtelomeric rearrangements because these regions are gene-rich and particularly prone to genomic instability due to the repeat sequences found in these areas [Linardopoulou et al., 2005; Riethman et al., 2005]. The most commonly used methods to detect subtelomeric rearrangements include GTG-banding, FISH with a complete set of subtelomeric probes (subtelomere-FISH), multiplex ligation-dependent probe amplification, and more recently microarray-based comparative genomic hybridization (array-CGH). The advantages and disadvantages of array-CGH over other assays have been described in detail elsewhere [Stankiewicz and Beaudet, 2007]. Array-CGH has become the method of choice for high-resolution genome analysis and screening for genomic copy number changes [Smeets, 2004; Cheung et al., 2005; Le Caignec et al., 2005; Sanlaville et al., 2005; Bar-Shira et al., 2006; Poss et al., 2006; Shaffer and Bejjani, 2006; Lu et al., 2007]. Several groups evaluated patients with DD/MR using subtelomere specific arrays with coverage ranging from several hundred kb to 5 Mb of subtelomeric regions and showed a high sensitivity of such assays in detecting known and novel cytogenetic abnormalities [Veltman et al., 2002; Harada et al., 2004; Kok et al., 2005]. Whole genome or targeted array-CGH has also proven to be a powerful tool for the rapid and accurate detection of subtelomeric rearrangements [Le Caignec et al., 2005; Bar-Shira et al., 2006; Shaffer et al., 2006; Ballif et al., 2007; Fan et al., 2007; Lu et al., 2007; Martin et al., 2007].
Targeted array-CGH with extended coverage at subtelomeric regions was first introduced at Baylor College of Medicine in February of 2004 as a clinical test coupled with FISH verification [Cheung et al., 2005]. We have defined the chromosome ends up to 10 Mb as the subtelomeric regions. In this study, we applied chromosome microarray analysis (CMA) to investigate 5,380 consecutive clinical patients and identified 499 cases with subtelomeric imbalances (9.3%) of which 236 cases were interpreted as pathogenic chromosomal abnormalities (4.4%). Our results support the previous observations that array-CGH is a sensitive and robust platform for the detection of unbalanced subtelomeric abnormalities and revealed genomic gains consistent with duplications as a frequent cause of subtelomeric imbalances.
We analyzed the results obtained from 5,380 consecutive clinical cases submitted to the Kleberg Cytogenetics Laboratory of Baylor College of Medicine for CMA analysis between July 16, 2005 and December 31, 2006. We used CMA V5 for the first 4,493 patients and CMA V6 for the next 887 patients. These samples were submitted by a variety of health care providers, predominantly pediatricians, geneticists, and neurologists from the United States and abroad. Patients referred for CMA analysis had diverse clinical phenotypes, including but not limited to DD/MR, dysmorphic features (DF), MCA, seizure disorders (SD), and autistic or other behavioral abnormalities. Ages varied from 1 day to 75 years, with a median age of 4 years. Among these patients, 2,550 (47.4%) had a normal prior or concurrent karyotype analysis, 175 (3.2%) had an abnormal prior or concurrent karyotype analysis, the remaining 2,655 patients (49.3%) had no or unavailable karyotype analysis. Overall, 278 patients had a prior normal subtelomere-FISH analysis.
The CMA V5 array has 853 BAC clones, interrogating genomic regions for >70 genomic disorders [Lupski, 1998], 41 subtelomeric regions (average 10 Mb/region), and 43 pericentric regions (www.bcm.edu/cma/table.htm, Chip Map V5.0). Of the 853 distinct clones, 481 cover subtelomeric regions with an average density of 12 clones per region. The subsequent and enhanced CMA V6 array included 1,475 clones covering ~150 genomic disorders and backbone clones for every G-band at the 650-band level (www.bcm.edu/cma/table.htm, Chip Map V6.0). The subtelomeric region coverage in CMA V6 is similar to CMA V5 except for the removal of 17 clones and addition of 37 new clones yielding a total of 501 clones. The clones were selected from NCBI/UCSC public databases or from the FISH probe panel routinely used in our cytogenetic laboratory. All clones were FISH-verified before use on the microarray. Any clone that hybridized to multiple chromosomal locations or to a different chromosomal region was excluded from the array. In general, 3–5 BAC/PAC clones were selected to cover the most distal 1 Mb and 1 BAC/PAC clone was selected for each proximal Mb. The starting point of the subtelomere coverage from telomere ranged from 0.04 to 0.8 Mb with an average of 0.24 Mb. The clones were chosen from segmental duplication free regions or contain segmental duplication sequence less than 30% of the clone sequence. Versions of the array are periodically updated to add clones of clinical significance and remove clones found to be highly polymorphic.
The manufacture of arrays, experimental procedures and data analysis methods were described elsewhere [Cheung et al., 2005; Lu et al., 2007]. For each patient sample, two experiments were performed with a dye-swap for the test sample and gender-matched control. Data were subjected to normalization and the dye-reversed results were integrated to determine inferences for each clone. The log2 test/reference intensity ratio and T-statistics permutation based P-value were the parameters used to objectively determine whether a clone deviates significantly from the baseline. Abnormal results were classified as gain or loss. A gain was defined when the combined log2 ratio of a clone was greater than 0.2 and T-statistics permutation based P-value was <0.05 whereas a loss was defined by the combined log2 ratio <−0.2 and T-statistics permutation based P-value <0.05.
All potential cytogenetic alterations detected on the arrays were verified independently by GTG-banding and/or FISH tests except copy number polymorphisms which had been detected frequently in patients and observed in multiple unaffected parents. The confirmatory method used was dependent on the size of the genomic imbalance detected by CMA. Confirmatory FISH was performed with the BAC/PACs that showed the alterations on the micro-array. Metaphase and/or interphase preparations from patient peripheral lymphocytes and FISH analysis followed a standard clinical cytogenetics laboratory protocol [Trask, 1991]. We requested parental samples for patients with copy number changes of clinical significance or unknown significance. The parental samples were studied either by FISH with the clone altered in the proband or karyotype depending on the size of copy number change detected in the proband. If the copy number change in the proband does not exist in the father nor the mother, it will be interpreted as de novo. However, paternity or maternity tests were not regularly performed in the laboratory.
We used CMA V5 for the first 4,493 patients and CMA V6 for the next 887 patients. From the 5,380 consecutive clinical cases evaluated by CMA, we detected DNA copy number alterations in subtelomeric regions in 499 (9.3%) patients. In 175 patients (3.2%), the alterations were seen in both the proband and a normal parent and therefore interpreted as benign familial variants. In 88 patients (1.7%), the parental samples were not available and since the alterations were small (≤1 Mb) and did not involve a known disease region, the clinical relevance could not be determined. Most of the subtelomeric familial variants (98/175) and variants with unknown clinical relevance (54/88) were consistent with copy number variations (CNVs) as archived in the Human Genome Segmental Duplication Database (http://projects.tcag.ca/humandup). The majority of these variants, with the exception of 11 single-clone changes, have been reported in an earlier publication [Lu et al., 2007] and therefore are not discussed in this study.
In the remaining 236 patients (4.4%), the rearrangements were either de novo (41), or inherited from a parent with either a similar phenotype or a balanced translocation carrier (20), or occurred in known disease regions (20), or were greater than 1 Mb in size when the inheritance was unknown (155), and therefore were interpreted as pathogenic. Among these abnormalities, 85% (200/236) were identified by CMA V5 with an overall detection rate of 4.4% (200/4,493) and the remaining 15% (36/236) were identified by CMA V6 with an overall detection rate of 4.1% (36/887).
The most common indications for referral in our patients were DD/MR, DF, MCA, SD, and autism. The detection rates for subtelomeric rearrangements in patients with these different referral indications ranged from 1.1% to 5.7%, with the lowest observed in patients with autism and the highest in patients with DD/MR with DF and/or MCA (Table I). For patients with a simple deletion or duplication, patients with a more severe phenotype (DF, MCA) tend to have an increased percentage in the detected imbalances of >3 Mb as compared to those of ≤3 Mb. However, there was no significant difference between the percentage distribution of imbalances >3 Mb and that of ≤3 Mb based on a small sample size (Table II). Interestingly, among abnormalities detected in 24 patients submitted with a clinical indication of MCA, six patients submitted by six different clinicians had a deletion at 4q35 (four patients had a 4q35 deletion and two patients had an unbalanced translocation; Fig. 1). In other patient groups, subtelomeric imbalances were relatively distributed among chromosomes and any association between the referring phenotype and specific chromosomal abnormality was not readily apparent.
We grouped the clinically significant chromosomal imbalances in subtelomeric regions into three categories: (1) simple deletions (103 patients) or duplications (58 patients); (2) unbalanced translocations (44 patients); and (3) complex rearrangements (31 patients).
We also grouped our patients into three categories according to their previous or parallel cytogenetic analysis (Table III): 2,550 patients (47.4%) with normal karyotype analysis; (2) 175 patients (3.2%) had an abnormal karyotype analysis; (3) 2655 patients (49.3%) had no or unavailable karyotype analysis. Seventy-six subtelomeric rearrangements were found among patients with an abnormal karyotype analysis. Other chromosomal abnormalities were either beyond the subtelomeric regions or balanced rearrangements, therefore were not included in this paper or could not be identified by our targeted array-CGH, respectively. As anticipated for the array-CGH analysis, this was the group with the highest abnormality detection rate of 43.4%, among which were 31 simple deletions, four simple duplications, 23 unbalanced translocations, and 18 complex rearrangements. The smallest visible rearrangement was 2.5 Mb. The CMA results were concordant with the karyotype analysis. Eighty-four subtelomeric rearrangements (36 simple deletions, 24 duplications, 12 unbalanced translocations, and 12 complex rearrangements) were found among patients with no or unavailable karyotype analysis with a detection rate of 3.16%. The remaining 76 subtelomeric rearrangements (36 simple deletions, 30 simple duplications, nine unbalanced translocations, and one complex rearrangement) were found in patients with a normal karyotype analysis; the group with the lowest detection rate of 2.98%. Eight deletions, five duplications, and four unbalanced translocations with an estimated size ≥5 Mb were initially not detected by karyotype analysis, although in retrospect, some of them were apparent.
Two hundred seventy-eight patients had a previous normal subtelomere-FISH analysis. Among them six subtelomeric rearrangements were identified by CMA (an interstitial deletion at 1p36.3 and 10q26, a terminal duplication at 4q35, and three interstitial duplications at Xq28). As expected, a previously identified apparently balanced translocation and a pericentric inversion were not resolved by array-CGH. Two terminal deletions in the long arm of acrocentric chromosomes identified by array-CGH were found to involve a ring chromosome by karyotype analysis. A mosaic deletion identified in approximately 75% of cells was discerned by both array-CGH and concurrent karyotype analysis.
Figure 2 shows the chromosomal distribution by arm (p or q) for simple imbalances. The most commonly detected subtelomeric alterations were 1p36.3 deletion (15), 22q13.3 deletion (10), and Xq28 duplication involving MECP2 gene (16), comprising 25% of all simple subtelomeric imbalances.
Figure 3 shows the regions and estimated sizes of the simple deletions and duplications. The majority of deleted (91/103=88.3%) and duplicated (51/58=87.9%) genomic intervals were located in or extended from the terminal 5 Mb. The rearrangements were interstitial in 39/103 (37.9%) deletions and 44/58 (75.9%) duplications.
Figure 4 shows the cumulative percentage distribution of simple imbalances by size. Rearrangements ≤3 Mb in size constituted 42.7% (44/103) of the deletions and 79.3% (46/58) of the duplications, the majority of which would not be identified by karyotype analysis alone.
The detection rate of clinically significant chromosomal imbalances is ~9.6% using targeted array CMA V5 and ~13.5% for targeted array CMA V6 (unpublished data). Therefore, the abnormality rate detected of 4.4% (4.4% in V5 and 4.1% in V6) in subtelomeric regions in this patient cohort is about 46% of all the genomic abnormalities identified by CMA V5 and 30% of those identified by CMA V6. The subtelomeric regions are covered at a higher density than the average interstitial region in the genome; therefore, it is unclear whether subtelomeric regions are more frequently affected than interstitial genomic regions in this patient cohort.
The overall disease-related subtelomeric aberrations in patients with DD/MR has been estimated as 5%, ranging from 0% to 23% depending on the sample size and patient ascertainment criteria [Biesecker, 2002; Xu and Chen, 2003; Moog et al., 2005; Yu et al., 2005; Menten et al., 2006]. The detection rate is higher in patients with a severe phenotype than in patients with a milder phenotype [Knight et al., 1999]. Our detection rate of 4.4%, which reflects pathogenic subtelomeric rearrangements in unselected patients, regardless of the severity of DD/MR or presence of other indications, is significantly higher than that in the largest published study of similarly unselected patients (N=11,688 patients, detection rate=2.6%) using a subtelomere- FISH assay (Table IV, P≤0.01) [Ravnan et al., 2006]. The most likely reasons for this detection rate difference are that array-CGH has greater coverage at the subtelomeric regions than FISH (10 Mb vs. 200 kb) and duplications are difficult to identify using FISH analysis. Our detection rate is also higher than that in a recently published study using targeted array-CGH [Ballif et al., 2007]. The latter study had the similarly unselected patients regarding referral indication with the exception that only patients with no known cytogenetic abnormalities were included. Therefore, we excluded samples with known cytogenetic abnormalities from the current database and the detection rate for patients with a normal or no chromosome study at referral was 3.6%, still significantly higher than 2.4% (Table IV, P≤0.01). The main difference between the current study and that of Ballif et al. is that we reported 58 duplications as abnormal whereas none is reported in their study. The most commonly identified duplications in our patients were Xq28 duplications involving MECP2 gene (16/58). This may reflect our use of gender matched controls in the CGH versus the gender mismatched approaches of Ballif et al. (data not shown). The frequency of deletion in our study is slightly higher than their study (1.69% vs. 1.58%), however, it does not reach statistical significance. The frequencies of unbalanced translocation and complex rearrangements differ between the two studies, however, the combined frequency is similar (this study versus Ballif et al.: 0.92% vs. 0.85%), presumably the classification of complex rearrangements may be different between the two studies.
Contrary to what can be inferred from the literature, we found that duplications occurred frequently in the subtelomeric regions. The number of reported deletions is much higher than the number of duplications, which likely reflects both ascertainment bias and limitations of available diagnostic methods. Duplications may result in milder phenotypes, or certain characteristic features that are different to those of the corresponding deletions [Somerville et al., 2005; Berg et al., 2007; Kirchhoff et al., 2007], and consequently, patients may not present for clinical evaluation. In this patient cohort, the indication for referral for CMA is usually general and its value to reflect the detailed clinical phenotype is limited. Therefore, phenotypic consequence of gains versus losses for the same genomic interval is impossible to discern given the limited clinical information. In addition, duplication may evade detection by conventional cytogenetic techniques. The majority of subtelomeric duplications detected were ≤3 Mb in size and not detectable by chromosome analysis, and were interstitial in position and not detectable by subtelomere-FISH. With the wider application of microarray and whole genome scans, more duplication syndromes are likely to be defined.
In this study, the most commonly observed duplications were at Xq28 region. Xq28 deletion causing loss of function of MECP2 is associated with Rett syndrome in females. Duplications of this region causing increased dosage of MECP2 has been associated with neurodevelopmental delay in males [Van Esch et al., 2005; del Gaudio et al., 2006]. All patients (N=16) with an Xq28 duplication were male in this cohort. The increasing number of duplications at this region supports that it may be a common cause of DD in males. The size of duplication at Xq28 ranging from 0.8 to 2.5 Mb, is similar to the literature reports (0.2–2 Mb). The different sizes and breakpoints of Xq28 duplication in 5 of the 16 patients have been reported elsewhere [del Gaudio et al., 2006]. Variously oriented low copy repeats (LCRs) are found scattered throughout a ~4 Mb region flanking MECP2, suggesting the potential involvement of these LCRs stimulating the rearrangements [Van Esch et al., 2005; del Gaudio et al., 2006; Bauters et al., 2008].
One of the limitations of conventional cytogenetics testing is genomic resolution. At the usual 550-band level, the resolution is limited to 5–10 Mb. We detected 90 simple imbalances by array-CGH with a size ≤3 Mb, which would likely be missed by karyotype analysis. We also found subtelomeric imbalances in 76 patients with normal karyotypes results. As anticipated, array-CGH could not detect an apparently balanced translocation nor an inversion. Therefore, when a microarray result is normal and a chromosome anomaly is still suspected based on the clinical phenotype, a karyotype analysis should be considered. Array-CGH detects imbalances of CNVs, but provides no positional nor orientation information. Array-CGH may reveal a genomic change (i.e., duplication), but FISH analysis is necessary to determine if the duplicated genomic segment has been transposed to a new position. Targeted arrays have limited coverage in non-disease regions, challenging one to estimate the size of deletion or duplication when it only involves backbone clones due to the lack of clone coverage at the vicinity. High density whole genome oligonucleotide array has the capacity to detect more copy number changes and resolve smaller CNVs than high density whole genome BAC array [Aradhya et al., 2007]. With the evolving knowledge of CNVs, whole genome oligonucleotide arrays with higher density will eventually replace the BAC arrays and targeted arrays in clinical cytogenetics.
We observed the enrichment of 4q35 deletions in patients with MCA. Cytogenetically visible deletions of 4q31–q35 are associated with clefting craniofacial malformations and micrognathia [Brewer et al., 1998]. Small terminal deletions of 4q do not seem to lead to gross abnormalities and the predominant clinical feature is mild to moderate MR/learning disability and minor DF. One of the known disorders mapping to 4q35.1–q35.2 is facioscapulohumeral muscular dystrophy 1A (FSHMD1A), wherein the underlying defect is suggested to be a contraction of the subtelomeric macrosatellite repeat D4Z4 at 4q35 [Tupler and Gabellini, 2004]. Patients with 4q35 deletions do not have FSHMD symptoms [Tupler et al., 1996; Pickard et al., 2004]. Among our patients with 4q35 deletions, six had MCA with either DD/MR, DF, or SD, one had DF only, and one had respiratory distress syndrome and asthma exacerbation. Further study into the potential association of 4q35 deletion and MCA is ongoing.
A large fraction of copy number changes (175/499=35%) was interpreted as familial variants or with unknown clinical significance (88/499=18%), among which over half were consistent with the CNVs archived in public database. CNVs average 250 kb in size encompassing ~18% of the euchromatic genome, and overlap with genes frequently and may play a major role in phenotypic variation [Bechmann et al., 2007; Lupski, 2007; Scherer et al., 2007]. The interpretation of copy number changes is complicated by the frequent occurrence of CNVs in both patients and the normal population [Lee et al., 2007]. Familial subtelomeric abnormalities have been observed in over 50% of the families [Adeyinka et al., 2005], which makes parental studies essential in interpreting subtelomeric copy number changes. However, the lack of accessibility of family members for testing, incomplete penetrance, the clinical variability of chromosomal abnormalities, and multiple potential mechanisms (e.g., unmasking of recessive alleles) by which rearrangements may convey phenotypes [Lupski and Stankiewicz, 2005], are factors complicating interpretation. It will be possible to avoid segmental duplications with the application of oligonucleotide arrays into cytogenetics and interpret the contribution of CNVs to a phenotype using the evolving knowledge and expanding databases characterizing CNVs. It has been reported that ~6% of the CNVs in the apparently healthy population are >1 Mb in size [de Smith et al., 2007]. Caution should be exercised in explaining large copy number changes when the inheritance is unknown.
In conclusion, this study demonstrates that genomic imbalances at subtelomeric regions contribute significantly to congenital disorders. The application of array-CGH with extended coverage at subtelomeric regions will likely reveal more submicroscopic syndromes and particularly reveal novel duplication syndromes.
We thank all the referring physicians, patients and their families for their participation. We also thank all our colleagues and coworkers from the CMA lab, Array Production lab, Cytogenetic/FISH lab, and DNA Diagnostic lab for their contributions and support.