We read with great interest your recent publication by Lee et al. on the use of array comparative genomic hybridisation (CGH) for prenatal diagnosis: a cohort study of 3171 pregnancies.1 Such a large cohort is an important contribution to the existing literature.2 It is important then to compare and contrast the results of Lee’s study with those of other recently published cohorts.3,4

The abstract states that when array CGH is performed for a fetus with a congenital anomaly apparent on prenatal ultrasound, the test was effective in identifying submicroscopic genomic imbalances (i.e. that are unidentifiable by conventional G-band karyotyping) in 33 of 194 (17%) cases. However, on reviewing data within table 2 (and later on in the abstract) this rate is recorded as 16/194 (8.2%). We attempted to extract data on submicroscopic copy number changes from table 2, but there appeared to have been formatting errors (e.g. in the row ‘pathogenic karyotyping reports’ the columns appear to be shifted a space to the right).

It is unclear within the table how the calculated total of pathogenic karyotype and array CGH reports are constructed. We assume that pathogenic karyotype reports would be made up of those with numerical abnormality plus those with large chromosomal deletions and duplications, whereas pathogenic array reports would be made up of the former plus microscopic chromosomal deletions and duplications, plus variants of unknown significance. This, however, is not made clear. It is not certain whether there may have been instances where the array CGH test failed to identify a chromosomal anomaly detected by conventional karyotyping, and indeed whether both types of karyotype testing were performed for all fetuses included in the study. The authors state within the discussion that ‘when array CGH was performed alone no significantly pathological results were missed’, but without a reference test how could one be sure?

Two array CGH platforms were used within this study: a 1-Mb bacterial artificial chromosome (BAC) and a 60-kb oligonucleotide array. It appears that there was a switch to the higher resolution 60-kb array towards the end of the cohort study. However, table 3 is not transparent to which array was used to detect which abnormal result.

When extrapolating using the coordinates of the oligonucleotide probes in comparison with those of BAC probes aligned to the reference genome in ensembl, and in BlueGnome’s bluefuse multi database software, we predict that only one of the abnormal results detected by the higher resolution 60-kb array within Lee et al.’s cohort would have been missed by the 1-Mb BAC. This point should be commented on, as the lower resolution BAC would be predicted to produce less variants of unknown significance (VOUS), an important point to consider when using this technology on prenatal samples. The authors did not however state which array contributed to the five VOUS recorded.

Finally, the authors conclude that there is a 0.52% baseline risk of submicroscopic genomic imbalance, even in women with an uneventful prenatal ultrasound examination. The authors conclude that prenatal array CGH is a recommended additional test in pregnant women undergoing invasive testing, but do not comment on the additional VOUS results that such testing could reveal, and the problems of interpretation and counselling of these variants in the prenatal setting. Before we can think about using this technology, even as an adjunct to conventional karyotypic testing, this issue must be addressed and the additional costs to overall pregnancy health care must be considered.


  1. Top of page
  2. References