CAR T-cell Characteristics Determine Efficacy

A study of patients with large B-cell lymphoma (LBCL) identified cells in chimeric antigen receptor (CAR) T-cell infusions that may limit therapeutic responses or cause neurologic side effects (Nat Med 2020 Oct 5 [Epub ahead of print]). The findings suggest several options for boosting the efficacy of CAR T-cell infusions and reducing risks.

Many patients who receive anti-CD19 CAR T-cell therapy have lasting remissions, but more than half don't benefit. In addition, some patients develop a systemic inflammatory response, called cytokine release syndrome, or a more serious problem, called immune effector cell–associated neurotoxicity syndrome (ICANS), that can cause confusion, seizures, coma, and death.

To determine why responses are so diverse, Michael Green, PhD, of The University of Texas MD Anderson Cancer Center in Houston, and colleagues performed single-cell RNA sequencing of axicabtagene ciloleucel (Yescarta; Kite/Gilead) products infused into 24 patients with LBCL. They correlated the sequencing results with the patients' clinical responses 3 months later: 50% of them had progressive disease, 4% had a partial response, and 38% had complete responses. One patient died before being evaluated.

By analyzing the cell types in the infusions, the researchers found that central memory CD8+ T cells were more abundant in complete responders. These long-lived cells efficiently attack cancer cells, promoting durable responses. In contrast, patients whose infusion products had an abundance of exhausted CD8+ and CD4+ T cells were more likely to have partial responses or progressive disease. Exhausted T cells, Green says, “are incapable of doing their job.”

In patients with LBCL, tumor cells release large amounts of DNA fragments, recognizable because they carry distinctive mutations, and the relative abundance of these snippets can indicate tumor burden. The team sequenced cell-free DNA in blood samples taken 1 week, 2 weeks, and 1 month after the treatment. They found that 75% of patients who had complete responses showed at least a five-fold decrease in circulating tumor DNA (ctDNA) within a week of treatment. No patients with partial responses or progressive disease showed such a large reduction.

Green and colleagues also assessed CAR T cells' impact on side effects. Four patients developed grade 3 or 4 ICANS, and their infusions contained numerous cells with gene expression patterns characteristic of monocytes, the precursors of macrophages and dendritic cells. These cells produced cytokines, including IL1β that promote ICANS. However, monocytes shouldn't be present in CAR T-cell preparations, says Green, adding that “we went through that data hundreds of times” to confirm the results. The researchers are now working to identify the nature and origin of the cells.

“The findings here have the potential for making CAR T cells better and more effective,” says Peter Martin, MD, of Weill Cornell Medicine in New York, NY, who wasn't connected to the study. For instance, the researchers discovered that exhausted T cells often boosted expression of two inhibitory proteins, LAG3 and TIM3. Antibodies that block these proteins are entering clinical trials, and Green and colleagues are testing them in preclinical models of CAR T-cell therapy. In addition, drugs such as anakinra (Kineret; Sobi) that curtail IL1 signaling might prevent ICANS.

Another possible benefit from the results: Doctors might learn more quickly whether patients need different therapies. Standard procedure is to evaluate patients after 3 months, but Martin notes that the study indicates that testing ctDNA a week after an infusion may predict whether the treatment will work. –Mitch Leslie

For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.