Posted on October 5, 2022

Cancer Treatment Predictor May Not Work in Patients With African and Asian Ancestries

Rodrigo Perez Ortega, Science, October 4, 2022

A genetic signature widely used to guide cancer treatments may not work for patients with African and Asian ancestries, according to a new study. The finding could mean people with these backgrounds are receiving an expensive therapy that won’t help them and could even worsen their prognosis, scientists say.


Sequencing tumor DNA is now a part of routine cancer care. Identifying specific mutations can help physicians narrow down the best treatment. Clinical trials have shown tumors with many mutations—what’s known as a high tumor mutational burden (TMB)—respond well to drugs known as immune checkpoint inhibitors, which release a brake on immune cells and enable them to attack tumors.

A study from 2021 found a high TMB could predict whether patients would respond well to a type of immune checkpoint inhibitor known as pembrolizumab. {snip}

“It’s essentially one of the few FDA-approved pancancer biomarkers,” says Alexander Gusev, a statistical geneticist at the Dana-Farber Cancer Institute who led the new study.

But what determines whether a patient truly has a high TMB? Doctors typically compare a tumor’s genetic material against reference genomes found in huge databases that compile DNA from thousands of people. But these databases are not diverse, Gusev says. For instance, gnomAD is made up of 56,885 European sequences, but only 8128 African ones.

To find out whether this disparity matters, Gusev and his colleagues compared the tumor sequences from more than 3600 cancer samples with sequences from a reference panel, and with the patient’s own DNA taken from healthy cells. When compared with a reference panel, the tumors of people with European ancestry showed a 50% higher TMB than they actually had, the researchers found. For patients who didn’t have European ancestry, their estimated TMB was more than twice as large as their actual TMB, the team reported last week in Cancer Cell.

The disparity was largest in patients with African and Asian ancestry. {snip}

A common algorithm used to compare disease and reference genomes also led to more false positives among people of non-European descent. Only 21% of patients of European ancestry had false high-TMB misclassifications, compared with 37% and 44% of patients of Asian or African descent, respectively, the team found.