Genetics-first approach encouraged in diagnosing primary immune regulatory disorders

Key takeaways:

• Flow cytometry and antibody counts might not identify all patients.
• Genetic databases are available, but they require continuous updating.
• When testing is inconclusive, symptomology provides clues.

WASHINGTON — Genetic testing can provide answers about children who may have primary immunodeficiencies when antibody and cell testing meet their limits, according to a speaker at the Academy of Allergy, Asthma & Immunology Annual Meeting.

In fact, a genetics-first approach makes sense now for primary immune regulatory disorders (PIRDs), Kathleen Sullivan, MD, PhD, chief of the division of allergy and immunology, Children’s Hospital of Philadelphia (CHOP), said during her presentation.

“I’m not saying it’s necessarily the strategy of choice for [chronic granulomatous disease] or [severe combined immunodeficiency]. But for PIRDs, I think you can make an argument,” she said. “Maybe for PIRDs, you really want to lean hard into genetics.”

For example, Sullivan called insufficient cytotoxic T-lymphocyte antigen 4, or CTLA4, an archetypical PIRD, with clinical manifestations that are more penetrant than laboratory manifestations.

“If you use flow cytometry and antibodies as your main diagnostic approach, you’re going to miss patients,” Sullivan said. “That’s not to say you don’t want an IgG level, because maybe they need immunoglobulin replacement. But if you use that as your main diagnostic approach, you’re going to miss patients.”

There are similar issues with nuclear factor kappa B subunit 1 (NFKB1) and signal transducer and activator of transcription 1 (STAT1) gain of function, she continued.

“This is where this enthusiasm is coming from, for doing genetics early as your diagnostic approach,” Sullivan said.

Case study

As part of the Very Early Onset Inflammatory Bowel Disease (VEO-IBD) program at CHOP, Sullivan said that she uses genetics first.

“I actually do very little flow cytometry,” she said.

Sullivan and her colleagues have identified 10 genes that they expect cause VEO-IBD, along with 13 genes that have a known association and 10 genes that did not have any previous association with the disease. But only nine of the genes across these groups could provide “a hint of the diagnosis” via standard flow cytometry and immunoglobulins.

“You’re still going to need genetics, because for many of these, it’s just going to be low T cells that you find,” Sullivan said. “Flow cytometry and antibody testing, which we all love very much, will only take you so far, particularly in the land of PIRDs.”

Sullivan also described a case of an infant aged 15 months who had enteroviral meningoencephalitis at age 12 days and bloody stools at age 3 months. Also, there were no changes in these stools despite 15 formula changes over the next year.

“He was so malnourished, he actually couldn’t sit without assistance,” she said. “It was really heartbreaking to see this boy.”

The boy was anemic due to his blood loss, Sullivan continued, although she called his immunoglobulins “pretty good for a 15-month-old.” She also called his CD3 count of 95 “clearly awful, but he was so malnourished.”

A month of feeding brought his CD3 count up to 377, which Sullivan conceded wasn’t normal, but still better than 95. The B cells and NK cells were normal. The sequencing results for IRAK4, RAG1 and TLR3 indicated uncertain significance.

“This is what tangles us all up,” Sullivan said, noting that testing companies are not required to use criteria from the American College of Medical Genetics, which include pathogenic, likely pathogenic, variant of unknown significance (VUS), likely benign and benign.

“I have been at this meeting for not even 8 hours, and I’ve heard five people say, ‘I hate VUSs. VUSs are so hard,’” Sullivan said. “I hear you. I’m right there with you.”

For a gene variant to be considered pathogenic, Sullivan continued, there must be a published case saying that the variant caused disease.

“Likely pathogenic? It’s a variant that looks an awful lot like something that was published,” Sullivan said.

“Pathogenicity is dictated by published data right now. I think we will have better algorithms,” she said.

“But that means it’s on us as a community. We’ve got to publish our cases. And if it’s in a puny journal with a terrible impact factor, do it for the good of the specialty. It’s not going to enrich your CV necessarily, but it’s going to enrich us as a community.”

Benign variants are common among the population, Sullivan continued.

“People from non-European ancestry communities are really disadvantaged,” Sullivan said. “To get a benign classification, there has to be a lot of previously identified variants.”

Sullivan said that people from the Cape Verde Islands, for instance, may not have much representation in the databases, so they are less likely to see benign classifications and are more likely to see variants classified as VUS.

“These are problems that we all have to grapple with as a community,” she said.

Although IRAK4, RAG1 and TLR3 all had uncertain significance, Sullivan still needed answers. She noted that IRAK4 and TLR3 were heterozygous but RAG1 was homozygous.

“So, if you had to pick one to push on, it would be the RAG1,” she said. “So how could you evaluate that variant?”

Sullivan recommended going to the Genome Aggregation Database (gnomAD). Entering a gene into the database’s search field produces a constraint classification, although Sullivan warned that this is not particularly useful for autosomal recessive conditions.

“This constraint means how well does this gene tolerate variation,” she said. “You’re looking at the observed over expected ratio.”

Results under 0.35 indicate that the gene does not tolerate haplosufficiency and that it “hates variation,” Sullivan said.

The ratio for RAG1 was 0.53, showing that it tolerates variation.

“Many of the PIRDs are autosomal dominant. And so for the PIRDs, you do want to look at this constraint,” Sullivan said.

“If the observed over expected for a gene is, like, 0.1, then pretty much any amino acid change that you see is more likely to be causal than if the observed over expected is 0.7,” Sullivan said. “So do take a look at it.”

Also, gnomAD offers graphical representations of which gene variants are pathogenic, VUS or benign, with color coding indicating hotspots. Reporting on variants informs these representations. Based on this coding, RAG1 was likely pathogenic.

“Variant classifications change all the time, because new information has been added to the databases,” Sullivan said.

When sequencing for a patient is more than 6 months old, Sullivan said that she always goes through this exercise because it is a 2-minute way to make a diagnosis.

“If the variant has been reclassified from a VUS to benign or pathogenic, I don’t have any more work to do. I’ve already got my answer,” she said. “Do be aware that the classification does change all the time. So do try and remember to go through this exercise and just check.”

Since gnomAD depends on published cases, and it is impossible to publish every variant, Sullivan said, the AlphaMissense database is another option. It reports whether in silico amino acid changes have any impact on protein folding without using any clinical data. Also, AlphaMissense only reports on likelihood of pathogenicity.

“If an amino acid change changes the protein folding in a major way, it’s bright red, likely pathogenic,” Sullivan said. “If it changes the protein folding in a less dramatic way, it’s pink, likely pathogenic.”

Although this database has only been available for about 8 months, Sullivan said that she uses it “all the time” and that it has been accurate approximately 90% of the time in head-to-head analyses.

“So, it’s actually quite good,” she said.

Networking with other specialists is another option, Sullivan added.

“We are a small but mighty community, so always reach out to your friends and colleagues if you can,” she said.

Sullivan noted that Luigi D. Notarangelo, MD, an NIH Distinguished Investigator in the NIH’s immune deficiency genetics section, also has an assay to assess RAG function.

“We reached out to him,” Sullivan said. “This variant was hypomorphic in his assay.”

These indicators all suggest that the patient described earlier had an RAG1 deficiency, “but presented as a PIRD because patients do not read textbooks,” Sullivan said, adding that PIRDs are not limited to the list from the International Union of Immunological Societies.

“They can crop up in any other category as well,” she said.

Beyond testing

However, “genetics are just part of the story,” Sullivan said. “Don’t try and oversell it.”

Sullivan recalled that when genetic testing first became available, she told patients that she would sequence everything and get an answer.

“And that is just less often the case than any of us would like,” she said.

Sullivan advised clinicians to be methodical in reading reports, considering phenotypes, collecting additional data, and checking on redefinitions of pathogenicity. Also, asking other clinicians for their opinions is vital, she said.

“I do this all the time,” Sullivan said. “I get turned down maybe half the time, but half the time it’s a win. And that person is super excited to have a clinician reach out to them.”

Clinicians need to pay attention to their patients beyond the results of these tests, Sullivan added, especially since “gene variants hide all the time.”

“It is more important than ever to harness our clinical acumen,” Sullivan said.

“If you’re listening to the story, and the story screams out [autoimmune lymphoproliferative syndrome], big nodes, big spleen, big liver, lots of auto-antibodies, even if you don’t find the gene, you’ve got to respect the scream.”

Genetics are just data that require interpretation, Sullivan said.

“Sometimes sequencing gives you a diagnosis, and we all cheer when that happens. And when we do get a gene, it does inform the optimal treatment,” Sullivan said. “When you don’t get a clear genetic etiology, you still have to treat the patient.”

Despite the time and effort that interpreting these genetic results may require, Sullivan encouraged clinicians to follow through with them.

“I think sometimes people get a VUS report back and it just goes in the back of EPIC and no one ever looks at it again,” she said. “And that’s really doing, in some cases, a disservice to the patient.”

Sullivan also encouraged clinicians to persevere through insurance authorization because genetic testing is necessary.

“I know it’s awful. I hate it. I grouse about it, just like all of you. But it is the landscape that we live in,” she said.

Still, she remained optimistic about growth in genetic testing, as long as the community shares its data.

“The landscape is going to get better. We’re going to have more patients. I think the analytic algorithms are going to get better,” Sullivan said. “But right now, we need to support our specialty, and that means publishing cases.”