Chromosomal microarray (CMA) for autism: what it finds and what it misses

What a chromosomal microarray actually reads

Your child’s DNA is like a very long book, usually packaged into 46 chromosomes. Only a small part of it spells out instructions for building proteins; much of the rest sets when and how much of each instruction gets used.

A chromosomal microarray (CMA) doesn’t read the book sentence by sentence. It checks whether whole chapters are missing or repeated. In genetics terms, these missing or duplicated stretches of DNA are called copy-number variants, or CNVs. A deletion is a chapter that fell out. A duplication is a chapter that got printed twice by mistake.

This matters because losing or duplicating even a modest stretch of chromosome can remove or double up dozens of genes at once — genes involved in how the brain and body develop. A microarray is built to catch exactly this kind of large-scale change, quickly and across the whole genome in a single test.

What it is not built to catch: single-letter typos within a chapter. That’s a different kind of change, and a different test — more on that below.

A microarray cannot diagnose autism. No genetic test can. It can sometimes explain part of why your child’s development went the way it did, which is a different and narrower job.

Why it became a first-tier test

For a long time, the standard genetic test for a child with developmental differences was a karyotype — essentially a low-resolution photograph of the chromosomes, useful for spotting only the largest rearrangements. A 2010 consensus statement from the International Standard Cytogenomic Array Consortium, reviewing 33 studies and nearly 22,000 patients, found that chromosomal microarray finds a cause in 15–20% of people with unexplained developmental delay, intellectual disability, or autism — compared with about 3% for a standard karyotype, once well-known chromosomal syndromes like Down syndrome are set aside (Miller et al., 2010). That’s a large enough gap that the consortium recommended replacing karyotype with microarray as the first genetic test for this group of children.

A 2013 guideline from the American College of Medical Genetics and Genomics set out how a geneticist should approach autism specifically: confirm the autism diagnosis first, discuss testing options and their likely yield with the family before ordering anything, and build an evaluation plan around that individual child’s history and features rather than a fixed list (Schaefer & Mendelsohn, 2013). In practice this is why a microarray, often paired with fragile X testing, is what many families are offered first.

What it actually finds in autism

The 15–20% figure above comes from a broader group — developmental delay, intellectual disability, and autism combined, and it includes children referred because they were born with multiple physical differences, not only because of autism. When researchers looked specifically and only at autism, the number came in a bit lower.

A 2015 study in JAMA tested 258 unrelated autistic children with a microarray and found a molecular cause in 24 of them — 9.3% (95% CI, 6.1–13.5%) (Tammimies et al., 2015). The same study also ran exome sequencing on 95 parent-child trios from that group and found a cause in 8.4% of those, with a combined yield of 15.8% once both tests were used together. Yield was meaningfully higher in children who also had other physical features alongside autism — major congenital differences and minor physical differences, which the study scored for every child — than in children whose only feature was autism itself.

A 2019 meta-analysis, using earlier chromosomal-microarray studies as its point of comparison, cited the same broader 15–20% range for developmental disorders generally (Srivastava et al., 2019). Put together, the honest range a parent should hear is: about 1 in 11 (9.3%) across a whole clinic group of autistic children, lower for a child whose only feature is autism and higher for a child with other physical features alongside it, and about 1 in 6 (15.8%) when a microarray and exome sequencing are used together (Tammimies et al., 2015).

Ask your genetics team directly what is and isn’t known about any specific change found in your child — that detail matters more to your family than the general statistic above.

What it cannot see

A microarray is powerful for large deletions and duplications, and close to useless for several other kinds of genetic change that also cause autism and related conditions:

• Single-letter changes. A typo within a gene — one DNA letter swapped, added, or dropped — is invisible to a microarray. This is what exome or genome sequencing is built to catch, and it’s a large part of why sequencing finds a cause more often overall.
• Fragile X and other repeat expansions. Fragile X syndrome is caused by an expanded repeat within one gene, not a missing or duplicated chunk of chromosome. A microarray does not reliably detect it. Fragile X needs its own dedicated test — read more on the fragile X page.
• Balanced rearrangements. Sometimes a piece of chromosome moves to a new location, or two chromosomes swap segments, without any DNA actually being lost or gained. Because nothing is missing or duplicated, a microarray usually can’t see it, even though the rearrangement occasionally still causes problems by landing in the middle of an important gene. The 2010 consensus statement notes that these, along with low-level mosaicism, are real but relatively infrequent causes of the pattern a microarray is used to investigate — under 1% (Miller et al., 2010).
• Most regulatory DNA. Much of the genome doesn’t code for a protein directly but still controls when and how much of a gene gets used. A microarray can see a deletion or duplication that removes a stretch of this DNA, but neither a microarray nor exome sequencing reads single-letter changes in it. Only whole genome sequencing reads this DNA in any detail, and even then interpretation is still developing.

None of this means a microarray is a weak test — it does its specific job well. It means a normal microarray result answers one question and leaves several others open.

A microarray can also occasionally turn up something you weren’t looking for. A SNP-based microarray can show long stretches where both copies of a chromosome match unusually closely, which can point to related parents or to both copies of a chromosome coming from one parent rather than one from each. It can also, on rare occasion, flag a change linked to a condition that shows up in adulthood rather than childhood. Ask the lab up front what kinds of findings they report beyond the reason you tested, and what you can choose not to receive.

The five result classes

Every CNV a microarray finds gets sorted into one of five categories, under a joint standard from the American College of Medical Genetics and Genomics and the Clinical Genome Resource (Riggs et al., 2020):

1. Pathogenic — strong evidence this specific change causes the condition it’s associated with.
2. Likely pathogenic — evidence points the same direction, but isn’t yet as strong.
3. Uncertain significance (a variant of uncertain significance, or VUS) — not enough evidence yet to call it either way. This is common, not a diagnosis, and not necessarily worrying.
4. Likely benign — evidence suggests this change doesn’t cause problems.
5. Benign — this change is well-documented as a normal variation, no different from the countless small differences everyone’s DNA carries.

The 2020 standard also makes a point worth repeating to any clinician who hands you a result without context: classifying a variant is meant to reflect the evidence about that variant itself, deliberately separated from any prediction about what it means for your specific child’s future (Riggs et al., 2020). A VUS, or even a likely pathogenic result, is not a forecast. It’s a starting point for a conversation. A full walkthrough of what each class means for next steps is on the genetic testing results page.

When sequencing replaces or follows a microarray

Genetics practice has been moving. A 2019 meta-analysis and multidisciplinary consensus statement found that exome sequencing finds a cause in 36% of children with a neurodevelopmental difference — a group the authors define as global developmental delay, intellectual disability, and autism — markedly more than the 15–20% reported for microarray. The figure was 31% when the developmental difference was the only finding and 53% when other conditions were present. The authors proposed exome sequencing as the new first-tier test for this group (Srivastava et al., 2019). A 2021 evidence-based guideline from the American College of Medical Genetics and Genomics made a formal recommendation: exome or genome sequencing should be considered as a first- or second-tier test for children with congenital anomalies, developmental delay, or intellectual disability (Manickam et al., 2021).

In practice, this shift is uneven. Some clinics now start with sequencing. Others still start with a microarray, often because it’s faster to authorize with insurance or more established in their workflow, and add sequencing later if the microarray doesn’t explain enough. Neither order is wrong on its own — what matters is that your family knows which test has been done, which hasn’t, and what each one is actually capable of finding. The full comparison, including what a whole exome sequencing result can add, is on the whole exome sequencing page; the genetic testing overview walks through the whole decision from the start.

Sample, turnaround, and cost

A microarray is usually run from a blood draw, though some labs offer a saliva or cheek-swab kit, which can be easier for a young child. Turnaround time varies by lab; ask your clinic for their typical window rather than assuming. Ask at the appointment when you should expect to hear, and ask who to call if that date passes. If you have heard nothing and no one gave you a date, call the ordering clinic rather than waiting.

Cost and insurance coverage vary widely by state, plan, and lab, and we won’t guess at a number here. What consistently helps: ask your clinician for a letter of medical necessity before the test is ordered, and ask your insurer directly what a microarray costs under your plan. The genetic testing cost page walks through what to ask and how to plan for it.

Questions to ask genetics about a microarray result

1. What specific copy-number variant did you find, if any, and what class is it — pathogenic, likely pathogenic, uncertain, likely benign, or benign?
2. Has fragile X been tested separately, or only ruled out by inference?
3. Given this result, would you recommend exome or genome sequencing next — and why or why not?
4. If the result is a VUS, what would change the classification, and is re-analysis planned as more is learned?
5. Does this result suggest a specific screening schedule for anything besides autism — heart, kidney, hearing, vision, seizures?
6. Should other family members be tested, and what would a positive or negative result in a parent mean for future children?

Write the answers down. You will be asked the same questions again by the next specialist, the school, or your own memory at 2am.

Questions parents ask after reading this

1. Our geneticist mentioned “first-tier” and “second-tier” tests — what do those terms actually mean for what happens to our child next?
2. The result mentions a duplication no one has explained clearly. What do we do with that?
3. Is it worth paying out of pocket for exome sequencing if the microarray found nothing?
4. Our clinician wants to test us as parents too. Why, and do we have to?

Bring this list to your next genetics appointment. A specific question gets a specific, useful answer; a vague one usually doesn’t.

If you only do three things

1. Ask directly whether fragile X testing has been done separately — a microarray does not cover it.
2. If your result includes a VUS or any variant you don’t fully understand, ask genetics to explain it in one plain sentence, and write that sentence down.
3. Keep a copy of the result and its lab report somewhere you and your pediatrician can both find again — you will be asked for it more than once.

A microarray answers one specific, useful question. It was never meant to answer all of them.