Successful CAR T-Cell Tx Needs Good T Cells to Begin With

Alexander M. Castellino, PhD

March 20, 2018

Chimeric antigen receptor (CAR) T-cell therapy has produced some dramatic responses in pediatric acute lymphoblastic leukemia (ALL), but the responses have been less spectacular in other hematologic malignancies, and they have been disappointing in solid tumors.

The reason may lie in the starting material, according to new data discussed at a presscast held ahead of the American Association for Cancer Research (AACR) 2018 Annual Meeting.

Sharing some of the data that he will be presenting at the meeting, presenter David M. Barrett, MD, PhD, assistant professor of pediatrics at Children's Hospital of Philadelphia (CHOP), in Pennsylvania, said, "Starting with poor-quality T cells may be a key problem in manufacturing effective CAR T cells."

Barrett and colleagues prospectively characterized peripheral blood T cells from 157 pediatric cancer patients, who included patients with ALL, non-Hodgkin's lymphoma, neuroblastoma, osteosarcoma, rhabdomyosarcoma, Wilms' tumor, Hodgkin's lymphoma, chronic myelogenous leukemia, and Ewing's sarcoma.

The team assessed T cells at the time of diagnosis and after each cycle of chemotherapy.

Barrett explained that one of the challenges of treating children with CAR T cells is to make a functional CAR T-cell product. There are several steps in making a CAR T cell, he explained. Although manufacturing and quality-control steps are similar in all institutions, the starting material is beyond the control of the manufacturer. "T cells are obtained from the patient through pheresis, and we have no control over their quality," he emphasized.

"However, we do know what a high-quality CAR T cell looks like," Barrett said.

"We were interested in learning what made a good starting material for CAR manufacture, rather than repeat prior work on learning what a good CAR T cell looks like after it has been made," he said in an AACR release.

Barrett and his team at CHOP noticed that in a number of patients who did not respond to treatment, the difference was in the starting material. He showed that good-quality T cells show a fivefold expansion (passage) in response to CD3/CD28 exposure for 7 days — the method used at CHOP for expanding the T-cell population before they are engineered with the CAR.

"For a child who did not respond, the T cells died in the lab," he said.

Looking at T Cells

The team determined the quality of T cells obtained from patients first at diagnosis and then after each round of chemotherapy.

Using a small-scale approach that simulated CAR T-cell manufacturing, Barrett showed that at diagnosis and before chemotherapy, they could obtain good-quality T cells for patients with standard-risk and high-risk ALL and Wilms' tumor, but not for patients with the other cancers.

"More than 90% of patients [with ALL] had high-quality T cells at diagnosis, and that may have played a role into why pediatric ALL is one of the great successes of CAR T-cell therapy," Barrett said. "We may have been using uniquely well-suited, good material to start with," he added.

Barret related the experience at CHOP, where they have been successful in treating only three children with lymphomas, but more than 200 children with leukemia.

The difference is reflected in response rates. For adult patients with lymphoma who were treated with CAR T cells, the response rate was 50%, whereas the response rate was with more than 90% for pediatric patients with leukemia.

"There may be a fundamental problem with making CAR T cells from the T cells we collect from patients," he said.

Barrett noted that the children with Wilms' tumor provided T cells that performed well. He said that although they do not at present have a CAR T-cell product for Wilms' tumor, there is good reason to believe that the starting material will provide them with good-quality, functional CAR T cells that could be returned to a patient.

Barrett reported one factor that differentiated poor-quality T cells from good-quality ones. He showed nanostring metabolic panels of the T cells from patients and from normal donors. Good-quality T cells had metabolic profiles similar to those of normal donors and clustered together, whereas poor-quality T cells had a different metabolic profile. In poorly performing samples, the profiles have a bias toward cells that use glucose as their fuel source (the glycolytic pathway), whereas good-quality T cells use fatty acids as their fuel.

Barrett further showed that in a patient with ALL, each cycle of chemotherapy altered the metabolic profile of the T cells — from the fatty acid profile to the glycolytic profile.

Exposure particularly to cyclophosphamide and doxorubicin-containing regimens, which are associated with mitochondrial dysfunction, severely deplete CAR T-cell potential in patients.

"Cumulative chemotherapy is altering the metabolic profile of the T cells we collect and gradually turning them — by cycle 6 — into something that doesn't work anymore," Barrett said.

AACR President Michael A. Caliguiri, MD, PhD, who was one of the session moderators, said, "This is a great example of the intersection of immunotherapy, immuno-oncology, and precision medicine." He explained that the investigators used expression profiling tools in the context of immunotherapy to discover what is wrong and how it can perhaps be fixed.

Caliguiri further elaborated that in this instance, Barrett and his colleagues had identified a broad metabolic pathway that is associated with dysfunctional T cells. Identifying these T cells makes it possible to predict those patients who are going to do well and those who are not, he noted. Further, strategies to decrease the injury to T cells and reverse the pathway associated with the injury will make it possible to generate better-quality T cells, which will provide good outcomes with immunotherapy.

Implications for the Clinic

In the question-and-answer session, Barrett elaborated on how these data have already altered clinical practice for collecting T cells at CHOP. For pediatric patients, for patients with high-risk disease and poor cytogenetics, and for patients who have high minimal residual disease after induction, "we know that cumulative chemotherapy is progressively deteriorating the likelihood that these cells will make a functional CAR product," he explained. "That is why in these patients we collect T cells early, even if patients are not eligible for CAR T-cell trials," he noted.

For adult patients, Barrett advocated taking a hard look at the quality of T cells and go the route of adaptive manufacturing. "In the nature of personalized medicine, take a highly personal therapy and further personalize it so that each patient has a specific manufacturing process for CAR T cells tailored to whatever problem we may find in the patient," he further elaborated.

When asked whether CAR T-cell therapy should be used before chemotherapy, Barrett said that the answer depends on which cancer is being discussed. "In pediatric leukemia, 80% to 90% of patients are cured with standard chemotherapy, and these patients are lifelong disease free," he said. The bar to move CAR T-cell therapy in front of standard therapy for this disease is different from metastatic osteosarcoma, in which long-term survival with standard chemotherapy and surgical approaches is poor (~20%), he noted.

With respect to changing the metabolic profile of poor-quality T cells, Barrett indicated that it was possible to do just that by force-feeding cells different fuels. CD3/CD28 stimulation is the method typically used to expand T cells. "We are force-feeding cells with fatty acids, and others are providing neutral amino acids, such as arginine, as attractive alternate fuel sources so that cells do not use glucose as their fuel source," he said.

Other ways of drilling down into the pathways cells use is through the use of small molecule inhibitors and receptor blockers, such as blockers of the glucose transporter. It is a method to trick T cells into using a fuel of one's choice, he noted.

Barrett also noted that a marker that is 100% predictive of what makes a good CAR T cell is lacking. "Failure to expand is highly correlative and is much better at predicting who is going to fail than who is going to succeed," he said. "Passage is not always a good predictor of an active product," he added.

The investigators have disclosed no relevant financial relationships.

American Association for Cancer Research (AACR) 2018 Annual Meeting. Abstract 1631, to be presented on April 16, 2018.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.