NEW YORK – Using circulating tumor DNA sequencing and immunohistochemistry to analyze samples from patients with relapsed or refractory large B-cell lymphoma (LBCL) who received CD-19 CAR T cells, researchers have pinpointed CD58 mutations as a potential resistance biomarker that hinders long-term treatment benefit.
Researchers led by Robbie Majzner, an assistant professor of pediatrics at Stanford University's hematology and oncology division, described the experiments that led to this finding at the American Society of Hematology's annual meeting. Specifically, Majzner and colleagues established that the interaction between CD58 on tumor cells and its ligand on CD2 on T cells is critical for patients' ability to respond to CD19 CAR T-cell therapy. "We're really excited to have found this novel resistance mechanism, and we were really surprised how clear cut it was," said Majzner.
Using these biological insights, Majzner's group was also able to engineer chimeric antigen receptors (CARs) to integrate CD2 signaling in a manner that re-established treatment efficacy in tumor cells with CD58 mutations or loss, leading to durable tumor suppression in preclinical experiments. The data presented "suggest an opportunity to personalize immunotherapy for patients with large B-cell lymphoma," Catherine Bollard, director of the Center for Cancer and Immunology Research at the Children's National Research Institute in Washington, DC, said in introducing the presentation during a press briefing at the meeting.
Majzner said that his group is hoping to advance a next-generation CAR T-cell therapy candidate into the clinic that can overcome CD58 loss by 2022. While this investigation also suggests that CD58 mutation and expression analysis may be useful as a companion diagnostic to guide treatment strategy, Majzner said that more research is needed before such a test can be used to decide which patients should or should not receive CD19 CAR T cells.
Between 40 percent and 50 percent of LBCL patients receiving CD19 CAR T-cell therapies, such as axicabtagene ciloleucel (Gilead/Kite Pharma's Yescarta) and tisagenlecleucel (Novartis' Kymriah), experience durable responses. However, around a quarter of patients have very poor outcomes with a median overall survival of around six months.
"These patients almost all die," said Majzner at a press conference. "But because we know CAR T cells can induce long-term remission in patients, we felt that if we can just get that initial response rate up, find the mechanisms of resistance, and engineer around them, we really could cure more patients."
Before they could engineer a next-generation CAR that can overcome resistance, Majzner and his colleagues had to find a biological explanation for why these patients relapsed on CD19 CAR T cells. His team sequenced the ctDNA of 51 patients treated with axicabtagene ciloleucel (called axi-cel for short) at Stanford using CAPP-Seq and also evaluated their samples using IHC. To do the CAPP-seq analysis, researchers worked with Ash Alizadeh's lab at Stanford, where this blood-based, next-generation sequencing-based method of measuring cell-free tumor DNA was developed a few years ago.
Based on the CAPP-Seq analysis, researchers found that none of the patients with a CD58 mutation had a long-term response to the CAR T-cell therapy. Some patients had CD58 mutations at baseline, while in others, mutations emerged during treatment, leading to disease relapse.
Researchers further evaluated the role of CD58 expression using IHC and saw a similar association between CD58 loss and axi-cel resistance. And when Majzner and colleagues combined the data on 12 patients with either a CD58 mutation or lack of CD58 expression, it further suggested that patients with CD58 loss "almost never have long-term, durable complete responses to treatment with axi-cel," Majzner said, "whereas patients with wildtype CD58 really hold up to what we've seen … with about a 50 percent cure rate."
Median progression-free survival was three months for patients with a CD58 aberration on axi-cel but not reached for those with normally functioning CD58. Only one out of the dozen patients with a CD58 aberration achieved a durable response, while the other 11 patients experienced disease progression after initially responding. Patients with intact CD58 were more likely to have complete responses, while those with aberrant CD58 were likely to see partial responses. Specifically, 82 percent of patients with intact CD58 and 25 percent with aberrant CD58 had a complete response, and 58 percent of patients with CD58 mutations or loss versus 10 percent of patients with wildtype CD58 had partial responses.
Majzner and colleagues next turned to cell lines and mouse models to further understand the mechanisms by which CD58 limits response to CAR T cells. For example, in a B-cell line exposed to CD19 CAR T cells, when researchers knocked out CD58 function on cancer cells using CRISPR, the CAR T cells lost the ability to kill them, Majzner said.
Using this information to improve next-generation CARs proved more challenging. Researchers noted that when CD58 is mutated on cancer cells, it cannot interact with its ligand CD2 on T cells, which is necessary for co-stimulating T cells to recognize and attack tumors. Through further cell-based experiments, researchers demonstrated that CD2 co-stimulation drives CAR signaling and protein phosphorylation needed to kill tumor cells.
"This data is very surprising to us [in terms of] how important CD2 co-stimulation turned out to be for CAR T cells," Majzner said. "We always considered CAR T cells a one-shot package, or that they're already endowed with co-stimulation. This tells us that although they are endowed with co-stimulation, other co-stimulants on tumor cells really matter and can really drive that T-cell function."
In designing a next-generation CAR T-cell treatment that could overcome this resistance mechanism, Majzner's group first tried to integrate CD2 signaling into a CD22 CAR structure, and initially these treatments appeared to re-establish tumor cell killing in CD58-knockout cell lines. But when researchers investigated the activity of these re-engineered CAR T cells in vivo, they saw better disease control but eventual relapse. The CAR T cells "were unable to finish the job," Majzner said. "The CD58 knockout cells eventually grew out and we ended up with no prolongation of survival."
But then researchers considered how CD2 receptors actually work in the T cell and recognized that the receptors are provided to the T cell essentially as a separate molecule expressed on the cell surface. CARs try to crudely mimic the properties of T-cell receptors, Majzner explained in an email.
T-cell receptors receive "co-stimulation from molecules such as CD2 and CD28 in trans," he said, "basically as a separate molecule that is also natively expressed on the cell surface." However, "current generation CARs receive co-stimulation in cis," as part of the actual molecule, integrated into the signaling domains. This cis co-stimulation works for some receptors, such as CD28 and 4-1BB, but not for CD2, he noted, adding that in the research presented at the meeting, his group showed that CD2 "needs to be provided as a separate molecule (in trans) to the CAR, much like it is provided to the TCR," he noted.
"Only when we adopted this trans set up … did we have significant tumor control of the CD58 knockout tumor cells and significant prolongation of survival," Majzner said at the meeting.
"The elegant and important work" by Majzner and colleagues "suggests the importance of CD58-CD2 signaling towards the optimal efficacy of CAR-T function in vivo," said Josh Brody, director of the lymphoma immunotherapy program at Mount Sinai's Icahn School of Medicine. Brody cited the work of University of Manchester's Eleanor Cheadle and colleagues, who also explored the importance of endogenous CD2 signaling to the function of CD19 CAR T cells in a 2012 Gene Therapy paper, and noted that the resistance mechanism described by Majzner's group is based on a "well-validated concept."
As to the Majzner's proposed manner for engineering CD2 co-stimulation into a CAR, Brody is of the view that both cis and trans approaches could work if optimized. "Though the trans integration of CD2 into next-generation CAR T [cells] was more effective than cis integration, that may be more a result of implementation than inherent superiority of the trans approach," he said. "It is unclear whether the cis integration may have put CD2 signaling domains in a poor spatial 'fit' relative to CD3 or other signaling domains."
Still, Majzner and colleagues have engineered a traditional CD22 CAR and transduced it with a chimeric receptor that has a different specificity for CD19 and an inter-cellular domain that provides CD2 signaling in trans to the CD22 CAR. Although the researchers haven't yet decided whether this specific CAR will go into human trials, Majzner's group hopes to have a CAR T-cell therapy that overcomes CD58 mutations or loss in the clinic within 18 months.
As more CAR T cells are used to treat cancer patients, Majzner predicted that the poor response associated with CD58 will be seen more often. "We think this will end up being important in other malignancies as well, because CD58 mutations are common in other cancers, including myeloma, Hodgkin's lymphoma, and some solid tumors," he said.
He noted that although CD58 expression and mutation status may have CDx potential, the current research should not be used just yet to decide which patients should or should not get CD19 CAR T-cell therapies. "We have found a significant correlation between CD58 loss/mutations and a lack of a durable response in a single center series of about 50 patients," he said, adding that he hopes other researchers will start looking at the impact of CD58 aberrations on CD19 CAR T-cell response.
Brody observed that CD58 could be an "important biomarker" for predicting efficacy of CAR T-cell treatments. Moreover, he noted that commercially available NGS platforms, such as Foundation Medicine's FoundationOne Heme, can already gauge CD58 alterations. As such, assessing this biomarker "would not require additional assay development," he said. "Other groups can query their databases of which patients were CD58 mutated already and confirm (or refute) the findings of this excellent study."
Majzner would like to see the predictive role of CD58 confirmed prospectively in a larger series of patients from many centers. "Only then would I start thinking about screening for CD58 mutations when considering therapies," he said. "Of course, even then, biology is complicated, and I would not preclude a patient from receiving the therapy. Even patients who experience a six-month complete or partial response, but relapse due to CD58 mutations, would have still derived significant benefit from [CAR T-cell] therapy."