Chromosomal microarray analysis (CMA) is used to analyse the human genome in order to detect copy number variations (CNVs), which can represent the cause of an abnormal phenotype. CNVs are not always pathological, thus constituting a challenge when interpreting CMA results.1,2
Lee et al. report on a series of 3171 fetal DNA samples analysed with CMA for a wide range of heterogeneous indications, using different resolutions (1-Mb BAC microarray or 0.5 Mb oligonucleotide), regardless of indication.3 The study is significant, both in terms of sample size and results. However, the heterogeneity of both platform and indication renders the diagnostic utility of CMA in prenatal diagnosis problematic.
In spite of the authors’ claims, the data do not support the widespread and generalised use of CMA as a screening test. In cases investigated for advanced maternal age, arrays detected 12/1911 positive cases (0.6%). However, case 15 would have also been evident at karyotype, case 14 showed an unknown CNV, cases 21 and 23 presented a 17p12 microdeletion (hereditary neuropathy with liability to pressure palsies, HNPP, syndrome) and case 25 presented a microduplication of chromosome 22q11.2, the clinical phenotype of which is extremely variable. Furthermore, in the last three cases, the alterations were inherited, raising ethical dilemmas for genetic counselling. Thus, the diagnostic value of CMA in this cohort of pregnancies should be adjusted to 0.36% (7/1911). Aneuploidies were only detected in 37 cases, but in 31 of these there was a pre-existing increased risk. A specific test (quantitative fluorescent polymerase chain reaction, QF-PCR, or multiplex ligation-dependent probe amplification, MLPA) would have yielded the same results, averting a laborious and costly technique, while also sparing couples ethical dilemmas related to the interpretation of CMA results. Cases investigated for parental anxiety similarly had a low detection rate and the identification of variants of uncertain clinical significance. The diagnostic value in this cohort should be adjusted to 0.2% (cases 27 and 28), not considering HNPP microdeletions (cases 21 and 23), 22q11.2 microduplication (case 25) and unknown CNVs (cases 30 and 31). Overall, the CMA detection rates over standard karyotype in the absence of specific indications should be adjusted to 0.31% (9/2909), with interpretation problems arising in 0.28% of cases (8/2909).
To conclude, the utility of CMA lies in selected contexts: ultrasound abnormalities with normal karyotype (detection rate in Lee et al. 11.5%), chromosomal aberration (22.2%) and supernumerary marker chromosomes (50%).2,4 CMA does not have a diagnostic advantage for indications not specifically suggestive of genomic pathologies. On the contrary, the detection of unknown CNVs can render genetic counselling extremely difficult. Nevertheless, even if the possibility to detect unexpected or clinically unknown CNVs is the same in any cohort (independent of indication), the high detection rate of array-based comparative genomic hybridization (aCGH) in high-risk pregnancies justifies its use. In selected cases and for specific indications, the use of a platform array at higher resolution than the one used by Lee et al. would be more useful for diagnosic purposes. Lastly, further characterisation of CMA results by means of fluorescence in situ hybridization (FISH) confirmation tests should always be performed to better assess their clinical significance for the current pregnancy, and for the risk of recurrence.5,6