Skip to main content
Premium Trial:

Request an Annual Quote

Epigenetic Signatures in Eye Predict Rare Tumor Aggressiveness, Treatment Response


NEW YORK – Methylation patterns found on cell-free DNA (cfDNA) in the eye can differentiate retinoblastomas into more or less aggressive subtypes and potentially predict the likelihood of treatment success, according to recently published research from the Children's Hospital of Los Angeles (CHLA).

In a study published last month in Nature Communications, researchers identified 294 genes directly regulated by DNA methylation that are implicated in tumor suppression and oncogenic pathways by examining tumor-derived cfDNA found in the aqueous humor, or the clear fluid in the front of the eye.

"Retinoblastoma is rare among cancers because it cannot be directly biopsied," said Jesse Berry, the study's first author and an ocular oncologist at CHLA.

The disease also lacks reliable ways to make a prognosis, leaving physicians somewhat in the dark as to which patients will respond best to therapy and how aggressively to treat. Failure to respond to therapy typically results in the surgical removal of the affected eye(s).

The current findings expand on Berry's prior research which had first raised the possibility of performing liquid biopsies in the eye after discovering retinoblastoma-related copy number variations in cfDNA found within the aqueous humor.

"But not all retinoblastoma cancers have these genomic alterations," Berry said. "So we started to think about what other potential biomarkers we could look at."

Mike Walsh, a geneticist and pediatric oncologist at Memorial Sloan Kettering Cancer Center, described the study as an "innovative approach to molecularly profile a tumor that is challenging to do so given material constraints."

"The study team uses a 'Sherlocking' approach which should motivate others to look for alternative means to prognosticate tumors," he said.

Although taking a liquid biopsy from the eye sounds likely to feel more invasive than most other liquid biopsies, Berry pointed out that chemotherapies for retinoblastoma must generally be injected into the eyes, which involves withdrawing some of the aqueous humor.

"You have to take some of this fluid out anyway," she said, "and other centers were throwing it away, and we said, 'maybe there's something in there, and maybe we should study it.'"

Berry teamed up with Gangning Liang, an expert in epigenetics, to perform genome-scale DNA methylation profiling of cfDNA and their corresponding primary retinoblastoma tumors, integrating the results with existing retinoblastoma tumor DNA methylation profiles. 

After establishing that cfDNA methylation profiles reliably discriminated retinoblastomas from healthy samples, Berry, Liang, and their colleagues observed two main methylation pattern subsets corresponding to tumors that had responded to therapy and to those that didn't, suggesting that these subsets represented different degrees of tumor aggressiveness.

They also found hypermethylation of the RB1 promoter. The silencing of the RB1 gene that this causes is a known mechanism of nonheritable RB1 inactivation. The investigators found identical RB1 promoter methylation in both tumor tissue and aqueous humor from the same affected eye in one case, indicating that this biomarker may be reliably detected by an ocular liquid biopsy.

Finally, the CHLA team integrated methylation data with gene expression and performed pathway analysis to identify aberrantly methylated genes that could be involved in the occurrence and progression of retinoblastoma. In all, 294 genes appeared to be differentially regulated in retinoblastoma by aberrant DNA methylation.

"We have to validate this in a larger group," Berry said, "but it suggests that this methylation signature might tell you which eyes are bad players and likely to fail current treatments and which eyes are good players and likely to respond."

Berry and Liang see this study as one more step in a longer line of work. Although they now have a way to potentially estimate which retinoblastomas will be more aggressive, questions remain regarding whether and how a signature might change in the case of a relapse.

And in the face of any such change, Berry asks, "did the tumor go from less aggressive to more aggressive or vice versa? Is there something we could target at that time? What's the [molecular] driver of recurrence?"

Although Berry and Liang's research remains in relatively early stages, Berry said that they have received "a couple emails" from companies interested in working together.

The CHLA team is also not alone in pursuing this line of research. Already, groups at the Institute Curie in France, Birmingham Women's and Children's Hospitals in England, Sick Kids in Canada, and several other institutions are conducting similar investigations.

Berry pointed out that although RB was the first successfully cloned tumor suppressor, research into retinoblastoma subsequently lagged behind other cancers. Demonstrating that tumor information can be obtained from the aqueous humor, she said, has the potential to usher in a period of rapid catch-up.

"[Retinoblastoma] is rare," she said, "but we're talking about kids that get cancer in both eyes [and go] blind or are at risk of losing both eyes. They're at risk of dying if it's not treated, so if we can treat these kids better, I think it opens up a whole new world for us."