NEW YORK (GenomeWeb) - While chromosomal microarray analysis has become the first-tier cytogenetic test for diagnosing individuals with constitutional abnormalities around the world, geneticists continue to struggle with a high level of findings of unknown significance.
After conducting a review of more than a thousand cases analyzed by chromosomal microarray, a team of researchers from the University of Tartu in Estonia determined that while the adoption of CMA in some cases quintupled diagnostic yield compared to conventional approaches, 41 percent of those findings were of unclear significance.
The researchers described their study in a recent Molecular Genetics & Genomic Medicine paper.
"It seems that our ability to detect chromosomal abnormalities has far outpaced our ability to understand their role in disease," the researchers wrote. "The interpretation of CMA findings remains a rather difficult task requiring a close collaboration between clinicians and cytogeneticists."
Katrin Õunap, a professor in the department of pediatrics at the University of Tartu, told BioArray News this week that Tartu University Hospital has been ordering CMA for prenatal and postnatal cases since 2009, and that CMA has been offered as a first-tier cytogenetic test for postnatal patients with constitutional abnormalities in Estonia since 2011.
Between January 2009 and December 2012, 1,191 patients were analyzed using the technology, not an insignificant number in a country of 1.3 million people.
"We do this analysis in every case if there is clinical indication – developmental delay, congenital anomalies and/or autism spectrum disorder," said Õunap, a co-author on the new paper. "The optimal number is approximately 600 analyses per year in Estonia," she said. "There are approximately 14,000 live births in Estonia [annually] and approximately 4 percent is at risk for some kind of congenital or genetic anomaly."
Of the 1,191 patients screened, 90 percent had suspected developmental delay or intellectual disability, multiple congenital abnormalities, or autism spectrum disorders, and another 5 percent were family members of those affected. The remaining 5 percent consisted of 60 prenatal cases that were also analyzed by microarray. For prenatal CMA testing, the main indications were abnormal ultrasound findings, a positive aneuploidy screening result, family history of chromosomal abnormalities, and other conditions. Because of the relatively small prenatal cohort, the authors decided to avoid making any conclusions about applying CMA as a first-line test in prenatal diagnosis.
According to the paper, screening for chromosomal rearrangements was performed using Illumina HumanCNV370-Quad or HumanCytoSNP-12 BeadChips.
Abnormal results were reported in a quarter of patients, with a total of 351 findings, including 147 deletions, 106 duplications, 89 long contiguous stretches of homozygosity (LCSH) events, and nine aneuploidies. Of all of the categories, it was the LCSH events that the Estonian researchers struggled most to interpret, as they found it difficult to link the stretches of events to patients' phenotypes and, ultimately, most of the 41 percent of unclear findings were LCSH events.
The high proportion of unclear findings might have been due to the researchers' use of higher resolution microarrays, and Õunap acknowledged that the choice of platform may have had a role in producing such results, considering that other analyses had produced much lower proportions of unclear findings.
For example, researchers at Birmingham Women's Hospital in the UK last year discussed an 8 percent rate of unclear findings using a 60,000-probe Agilent Technologies' comparative genomic hybridization array, which is many times the 0.4 percent detection rate achieved using conventional cytogenetics in the same study.
"As the CMA platform resolution increases detection rates increase but are associated with an increase in VOUS rates," the British authors wrote in the paper.
The Estonians' microarray platforms in comparison offered an even higher resolution look at patients' genomes. Illumina's HumanCNV370-Quad contains about 370,000 SNPs, and its CytoSNP-12 includes about 300,000 markers.
"Many centers are using comparative genomic hybridization, which are not diagnosing [loss of heterozygosity] areas and [uniparental disomy]," Õunap noted.
Of the remaining findings, another 41 percent were defined as pathogenic or likely pathogenic, and 18 percent were classified as benign or likely benign. Clinically relevant findings were detected in 126 patients, the authors reported. Fluorescence in situ hybridization, quantitative PCR, G-banding, and multiplex ligation-dependent probe amplification were all used to confirm the results.
Altogether, implementation of CMA improved the diagnostic yield to between 10 and 25 percent per patient from a previous rate of between 5 and 6 percent detected with conventional cytogenetics. The improvement in yield matched the results of other studies. The International Standards for Cytogenomic Arrays consortium, for instance, published in 2010 a diagnostic yield of between 15 and 20 percent for CMA compared to a 3 percent yield for G-banded karyotyping.
While the authors acknowledged the similar results in the paper, they argued that the interpretation of CMA findings "remains a limiting factor hampering the selection of truly causative variants." Therefore, referring pediatricians or neurologists "should be aware of the possibility that CMA provides results which are often random or difficult to interpret," they wrote. The authors noted that the constant addition of information to open-access databases has made interpretation somewhat easier, as they were able to classify nearly a fifth of findings over time as benign or likely benign.
They also offered some caveats regarding array technology in general.
Commenting on platform selection, and given their own experience, the authors wrote that there is "no need for maximum resolution in a genome-wide clinical test, as this is accompanied with an increase in the number of findings with uncertain clinical significance."
And despite the high diagnostic yield achieved using CMA in clinical practice, they argued that conventional chromosome analysis and FISH "still remain useful tools for [characterizing] structural aberrations, because important disease mechanisms may go undiagnosed and may be underestimated if only CMA is performed."
In the paper, the authors noted that balanced translocations and inversions are not detectable by CMA.
Based on the review, Õunap said that it was unlikely her colleagues would alter the way in which they carry out CMA. However, she said that she expects the number of patients tested in Estonia to rise in the future. "We think that the optimal number for Estonia is 700 up to 750 CMA [cases] per year," she said.