Leukemic Cells in CAR-T Find a New Way to Evade Immune Recognition
Standard flow cytometry missed the detection of the CAR19 epitope on the surface of the leukemia cells.
Researchers from the Center for Cellular Immunotherapies, Perelman School of Medicine, the University of Pennsylvania, in Philadelphia (UPenn), led by chimeric antigen receptor t-cell therapy (CAR-T) pioneer Carl June, MD, released the results of a study that suggested that the inadvertent transduction of a single leukemic cell during the manufacturing process for tisagenlecleucel (Kymriah™) was the direct cause of a CAR19+ leukemia relapse in a patient being treated with the medication. The patient in question, patient #107, ultimately died from progressive leukemia, which investigators hypothesized was a direct result of the accidental transfer of chimeric antigen receptor (CAR)-transduced B cell leukemia (CARB) cells.
The culprit behind this relapse, investigators wrote in Nature Medicine, was the transfer of a single leukemic cell with “extensive replicative capacity” that snuck into the apheresis product while the medication was being manufactured.1 The UPenn researchers concluded that this was how the offending cancer cell infiltrated the gene-modified cell therapy — it passed through undetected during CTL019 production prior to infusion into the patient. This offending blast cell was found to be able to replicate both in culture and in vivo.
This is especially curious because the investigators confirmed the absence of detectable levels of leukemia cells at 28 days following infusion of the apheresis product, and the patient was in complete remission during this time, according to the researchers. However, once the investigators performed next-generation immunoglobulin heavy-chain sequencing (IgH-seq) on the patient's bone marrow retrospectively — but 3 months before the evidence of a clinical relapse even appeared — they were able to confirm that CAR transduction had occurred in leukemic cells.
Evidence of abnormal gene expression in leukemic cells, which the researchers called “ectopic CAR19 expression” on B cell acute lymphoblastic leukemia (B-ALL) cells, effectively created a CTL019-resistant blood cancer. As a result of not being fully sorted out of the collection of cells meant for reinfusion, leukemic cells remained in the apheresis product, and they became modified to express CAR receptors. These CAR receptors were able to mask CD19 and persist. And, in a validation experiment using a mouse model, researchers were able to confirm that the relapse cells that originated from a single blast clone were transduced with 2 copies of the CAR19 vector during drug production.
The malignant cells were not detected by the antibodies used in flow cytometry operations, the researchers reasoned, because all of the CD19-specific antibodies became bound in cis to the leukemic CAR19 and obfuscated recognition of the CD19 epitope by T cells. This conferred resistance to the action of CTL019 when the leukemic cells started to proliferate. Competitive binding assays confirmed that all CD19-specific monoclonal antibodies (mAbs) were blocked, even though immunohistochemical staining of the bone marrow at relapse revealed that CD19 protein expression was occurring on the surface of leukemia cells.
In fact, this may be the underlying reason for some of the relapses that have occurred following administration of these types of cell therapies. Some of the leukemic cells that were thought to have been teased out may slide through the filtration and cell selection processes undetected, armed with another invisibility cloak. Despite this, the researchers wrote that the mere presence of the rare leukemic blasts — found in 6 out of the 17 patients under investigation — did not always correlate with relapse. Instead, relapse was determined to be linked to the “higher frequency of leukemic B cell content in the initial apheresis” solution.1
The researchers confirmed that the relapse in patient #107 at month 9 was the direct result of the unintentional transfer of malignant cells by comparing the clonal relatedness of the cells used in the pre-CTL019 infusion product with those of the original leukemia. They found that the cells matched, suggesting the malignancy originated from one leukemic cell.
Despite the patient being treated with salvage therapy and 9 cycles of moxetumomab (an anti-CD22 antibody), and then with a CD22-directed CAR-T at the National Cancer Institute, the patient's CARB cells continued to expand until the patient's death. “Given that transduction of leukemia cells with a CD22 CAR leads to specific resistance to CD22 and not CD19 CARs,” the researchers wrote, “it is likely that this could be a general mechanism to render any tumor or normal cell specifically resistant to a CAR T cell.”1
When the researchers retrospectively analyzed the samples from patients enrolled in their previous CTL019 trials, they determined that it was unclear whether the 1 relapse they recorded previously was associated with the CD19 splicing pattern they observed with patient #107. Nonetheless, they noted that better cell separation methods prior to apheresis are warranted.
When asked how to ensure that all leukemic cells have been successfully filtered out of apheresis preparations, one of the study authors, Jan Joseph Melenhorst, PhD, director of the Product Development & Correlative Sciences Laboratory at UPenn, replied: "We will never be able to ensure that all leukemic cells have been removed from the cell product since we are only sampling a small aliquot of the final product. Bear in mind that this has not occurred before or after this discovery — at UPenn and elsewhere.”
“Since this event, though, we carefully track CAR-marked cells and will start an investigation whenever the dynamics of CAR-marked cells display a similar behavior as with this patient,” Dr Melenhorst added. “Going forward, however, we and others have improved our methods of selecting T cells at the start of CAR T cell manufacturing.”
And since the publication of this article, Dr Melenhorst and his team appeared to be confident that the transfer of malignant cells during the administration of CAR-T is not a widespread nor common event. “In this case, we had a single leukemic cell that was transduced and survived the culture. Both are very unusual events,” he added.
For its part, Novartis acknowledged the paper by Dr June and colleagues, and said the patient in question who was enrolled in the first CART19 trial was treated with a UPenn-manufactured CAR-T cell therapy. In a note to Cancer Therapy Advisor, Novartis Spokesperson Julie Masow wrote, "The Novartis manufacturing process is different from the academic manufacturing process used at Penn. We are not aware of any cases of this happening in the more than 400 patients treated with CTL019/Kymriah manufactured by Novartis for clinical trials or the commercial setting ... We have processes in place to eliminate B cells, including leukemic cells, from the apheresed product and have steps throughout the manufacturing process to further limit the risk of B-cell presence. Novartis also has a long-term registry to monitor Kymriah patients for 15 years following treatment."
- Ruella M, Xu J, Barrett DM, et al. Induction of resistance to chimeric antigen receptor T cell therapy by transduction of a single leukemic B cell[published online October 1, 2018]. Nat Med. doi: 10.1038/s41591-018-0201-9