In our retrospective study we failed to show that the detection rate of aCGH at a median resolution of 600 kb (400 kb respectively) is superior to conventional chromosome analysis. The detection rate of CNVs of unknown significance, discussed as a major hindrance to the introduction of aCGH in prenatal diagnosis, is at 2%. Due to the retrospective anonymous design of the study we were not able to confirm or exclude familial segregation to further elucidate pathogenicity.
In contrast to the existing reports focusing on detection rates in pregnancies with fetal malformations mostly diagnosed by second trimester sonography we dealt with pregnancies showing a high risk for chromosomal anomalies in the first trimester for heterogeneous clinical reasons. Clinical indications for invasive prenatal testing were increased nuchal translucency (>3 mm) and/or elevated first trimester test risk in 57, more than a half, of the referred samples. Recently, Leung et al. reported on the detection of 4 additional pathogenic submicroscopic anomalies ranging from 1.2 to 7.9 Mb when using a custom designed array at a median resolution of 100 kb in a series of 48 pregnancies with nuchal translucency more than 3.5 mm . However, 2 of these fetuses had additional sonographic anomalies. Both sample sizes, Leung’s and our’s, are comparably small and the overall detection rate of pathogenic CNVs in fetuses with increased NT seems to be rather low, but might nevertheless require further and larger studies. The progression or regression of cystic hygroma should probably be taken into account for phenotype correlations, and other monogenic disorders such as Noonan or CFC -syndrome caused by mutations in genes of the RAS-signalling pathway, seem to be a reasonable differential diagnosis to consider [22–24] (and personal communication Yntema H.G. European Society Human Genetics C07.6).
In our approach we have chosen a comparably low and targeted resolution of a median 600 kb resolution for the BAC-array in order to avoid CNVs of unknown clinical significance. In a prenatal setting unknown CNVs can represent a particular challenge as phenotypic data are sparse compared to the possibilities of obtaining phenotypic details on postnatal patients. In addition, the potential option of pregnancy termination requires a higher level of interpretation accuracy. Our resolution might be too cautious especially considering our low detection rate of pathogenic imbalances. Most postnatally identified pathogenic microdeletion- and duplication syndromes, however, are caused by pathogenic CNVs larger than 400 kb . Therefore, a whole genome resolution of 400 kb should provide a reliable prenatal diagnosis. After extensive individual counselling diagnostic array analysis is now offered as a diagnostic choice in high risk pregnancies with fetal malformations and normal conventional chromosome analysis lacking further diagnostic options or to determine the precise nature of a structural aberration detected by conventional karyotyping.
D’Amours et al. investigated a series of 49 fetuses with ultrasound anomalies using custom-designed whole genome oligo-arrays with different resolutions . Out of 4 clinically significant anomalies, all detected by high resolution arrays, only 1 was smaller than 400 kb, one was about 700 kb and 2 were potentially detectable by conventional chromosome analysis at sizes of about 8 Mb and 14 Mb. Recent high resolution array studies using heterogeneous array designs including BAC-, oligo- and SNP arrays also focusing on detection rates in fetuses with congenital malformations [12, 27–29] yielded variable detection rates of pathogenic CNVs in 1–3% up to 10% , and even up to 16% . However, sizes of the clinically relevant aberrations were larger than 1 Mb [12, 28, 29]. Even in the studies using whole genome SNP-arrays with a resolution higher than 100 kb clinically relevant aberrations were mainly in the range of 1 Mb or more  or larger than 500 kb or 1 Mb [14, 25] in all but one case . An overall detection rate of 5-6% also including chromosomal anomalies detectable by conventional karyotyping was shown in array studies dealing with samples of heterogeneous clinical indications including fetal structural anomalies [12, 30, 31]. Hillman et al. concludes an overall detection rate of 3.6% regardless of the clinical indication and 5.3% for cases with ultrasound anomalies in his review and meta-analysis of this recent heterogeneous literature .
Referring to these results we would have expected to find at least one additional submicrosopic anomaly in our cohort examined. We focused, however, on pregnancies at risk for chromosomal anomalies as determined by first trimester screening, which has not been systematically considered in previous studies. Our patient cohort likely represents heterogeneous clinical indications, the main group being referred for increased nuchal translucency and elevated first trimester risk screening, whereas the proportion of fetuses with congenital malformations is comparatively low. Our results are in accordance with the findings of Faas et al. (personal communication, ESHG 2011 P05.32) who did not find any clinically relevant abnormalities in a cohort of 95 fetal samples with ultrasound anomalies including increased nuchal translucency when implementing a strategy for routine array diagnosis in parallel to QF-PCR. High resolution whole genome arrays may therefore be specifically indicated in pregnancies with distinct fetal ultrasound anomalies or in the precise determination of rare structural chromosomal anomalies.
Our approach revealed a detection rate of 2% for CNVs of unknown significance at the given resolution, confirming that aCGH seems to be a safe approach in the diagnosis of first trimester pregnancies considered at high risk for chromosomal anomalies. In our experience aCGH is technically feasible and reliable as a first-line diagnostic test for prenatal samples, replacing laborious direct preparation of CVS as well as other rapid but less comprehensive testing procedures. We were able to provide a result in 3 to a maximum of 5 days after DNA extraction offering a superior coverage of pathogenic aberrations in the genome. Thus, aCGH provides the advantage to give patients a more complete genome analysis result in a short turn-around-time. High resolution arrays may gain further importance as we can expect that the clinical use of non-invasive diagnosis for common aneuploidies will increase the proportion of samples from high risk pregnancies obtained by invasive diagnosis. However, the appropriate array resolution still remains a matter of debate and may need to be adapted to the specific clinical indication and/or parental expectations. These would require a flexible resolution design of arrays and analysis software as well as the possibility to analyze one single patient in a (very) short turnaround time or multiple patients in parallel for diagnostic laboratories. Platforms requiring small initial DNA amount will be advantageous as the DNA amount from direct villi or amniocytes will be sufficient to allow array analysis as compared to time-consuming cell culturing for DNA extraction. For the time being cell cultures may be used as a back-up for confirmational FISH-analysis as well as follow-up of mosaic results if aCGH reveals an unbalanced structural aberration.