10 Common Myths About Microarrays

Microarrays or Chromosomal Microarray Analysis (CMA) have revolutionized the genomics research and applications. Like any technology which emerges in research and then gets adopted in various areas, Microarrays have also come a long way from being an experimental concept and to become a commonly used technique in a wide variety of applications for studying the genome.

Along the way, the technology has indeed been improved and has had several variations in terms of the assay design, processing, bioinformatics and interpretation. The variations have led to various products and platforms being available in the market and each proclaiming an improvement or a targeted application.

Over the years, given the variety of products and applications, certain myths have emerged with respect to utility of microarrays in the clinical diagnostic settings. Here are 10 common myths associated with microarrays and the explanations to dispel the same.

Like most new technologies, microarrays also were first developed and applied in research settings. However they entered the clinical diagnostics arena as early as 2003 or even earlier when there were several research papers published comparing the results from microarrays and other commonly used diagnostic technologies. Given the concurrent results and higher diagnostic yields using microarrays, the technology increasingly got adopted in clinical practice. Several associations including American College of Obstertics and Gynecology (ACOG), Society for Fetal Medicine, American College of Medical Genetics (ACMG) and several others have recommended Microarrays for diagnostic use. Today, it is the most preferred and/or recommended first line of testing for specific disorders and clinical situations.

The concept of validation is indeed applicable for any new technique. However, as mentioned earlier, use of Microarrays in diagnostics is now more than a decade old with several publications and endorsements from medical bodies. The results of Microarray tests do not require validation per se. On the contrary, Microarrays provide an important validation step and guide for other technologies.

Microarrays can detect all types of sub microscopic deletions and duplications. However, Microarrays cannot detect balanced translocation. The balanced translocation can only be picked through karyotyping. Plus karyotyping is simpler and cost effective for detection of aneuplodies and larger chromosomal deletions, duplications and rearrangements. Hence while Microarrays are indeed very powerful technique, Karyotyping is still relevant and applicable in clinical diagnostics

NGS and CMA are two fundamentally different approaches to analyze the genome. The NGS is focussed on Single Nuecolotide Variations (SNVs) and the Microarray is focussed on Copy Number Variations (CNVs).There are different applications for each technology.

While there are few overlaps and instances where both technologies could be used, the application areas are quite different and well defined. For e.g. if you are looking for point mutations or sequence changes, NGS would be the appropriate technology. On the other hand, in case of deletion/duplications CMA is more suitable. Hence both NGS and CMA complement each other.

All new molecular technologies generate vast amounts of data. These data needs to be cleaned, sorted, annotated and interpreted which is done through bioinformatics – a combination of biology, information technology and data sciences. Like any other field, there are a vast number of approaches and tools for bioinformatics. And like any computer program algorithm, the bioinformatics algorithms can be designed and used in a wide variety of ways. So there is no ‘standardization’ per-se. The appropriate bioinformatics approach and algorithm is a function of what you are looking for in the data.

The microarray platform designers do provide associated bioinformatics tools that can be used to read and analyze the data generated. However, these tools need to be used with skill and expertise to arrive at proper conclusions. Improper handling of bioinformatics output can lead to misleading or confusing results.

In the context of clinical diagnosis, the bioinformatics output is not directly usable. It needs to be interpreted and correlated with clinical indications, existing research/publications and the current knowledge base.

When microarrays technology was first developed, it improved on the concept of comparative genomic hybridization (CGH) which was a molecular-cytogenetic method of identifying copying number changes by comparing the sample DNA with a reference/control DNA. The initial microarrays were known as Array-CGH as they used the Microarray chips/platforms to perform the comparative genomic hybridization. Array-CGH revolutionized the genome wide detection of copy number changes.

The advent of SNP based microarray chips has gradually replaced the Array-CGH in the diagnostic arena. The main reason for this is the ability of SNP arrays to detect copy-neutral loss of heterozygosity (CN-LOH), which is caused by uniparental disomy. An SNP array has probes to cover both the CNVs as well as SNPs and hence provide much more information about the genome that the previous approaches.

BACs-on-Beads technology, using BAC (bacterial artificial chromosome) is useful for rapid detection of aneuploidies (missing or extra chromosomes) and few other micro-deletions. It offers a wider coverage as compared to traditional Fluorescent In-Situ Hybridization (FISH) or QF-PCR techniques. However, the BAC on Beads does not provide a genome wide coverage for detection of CNVs or SNPs or UPDs.

Microarrays (both arrayCGH and SNP array) offer a much wider coverage and sensitivity and have hence replaced the BAC on Beads technology in clinical practice.

Some of the researchers and clinicians find the enormous data generated by the arrays as intimidating. Plus the inexperience in interpretation and fear of finding unexpected results leads one to conclude that ‘less is more’. Or perhaps in the context of SNVs and NGS, there is a case for a more targeted and specific assay.

However, in the context of Clinical Diagnosis and CNVs, the more the information, the better it is. Based on experience and proper data interpretation, it is indeed possible to draw clear cut conclusions for any case. Higher the resolution of the array, higher is the confidence in terms of the coverage and lesser is the probability of missing out on any important CNVs.

Even if there are some variations of uncertain significance, it is more important that we don’t miss out something significant rather than worry about discovering things which are not significant. Hence more the resolution of the microarray, the better is the confidence.

The wider and detailed coverage offered by Microarrays indeed enables them to unearth much more information than other technologies. Given that we are still learning new things about the human genome, the findings in microarrays may at times have an uncertain significance.

However, it all depends on the laboratory and the person doing the clinical interpretation (refer Myth # 5). With enough experience, a lab can confidently rule out several possibilities and present a more simplified analysis and interpretation of the results thereby reducing the reporting of ‘variants of uncertain significance (VOUS) where it is not necessary. This way the results and counselling becomes simple for the referring clinicians and the patients.

Although, Microarrays is a indeed widely used and validated technology, there are a wide range of differences between the microarray platforms and chips offered by different technology companies such as Affymetrics, Illumina, Agilent to name a few. Again, while most microarray vendors are offering a competing product for various applications, there are finer differences which only the experienced users can appreciate. Similarly, there is no standardization per-se in the bio-informatics offered by various vendors so that the results may be directly compared. There are a variety of parameters in bioinformatics which may be used to sort and cure the data. It is upto the person doing interpretation to extract the relevant data from the larger data set generated using a microarray.

Further, the drawing of inferences from the data set is dependent on the quality of the referenced database. The publicly available databases are not finely curated or updated. The diagnostic labs hence also use their own internal database base generated over the years to identify and classify the unique, pathogenic and non-pathogenic findings.

———–

Microarray technology is indeed very powerful and offers great utility for clinical diagnosis. However, given the choices and complexities involved in microarrays as well as genomic technologies in general, it is advisable to work with experienced scientists and laboratories who understand the potential as well as the boundaries of various technologies.