Attendees at ASH 2017 Photo courtesy of ASH

ATLANTA—Data suggest a therapeutic window may exist for chimeric antigen receptor (CAR) T-cell expansion with JCAR017, according to a preliminary model.

In a core set of 67 patients with diffuse large B-cell lymphoma (DLBCL) who had received JCAR017 in the TRANSCEND NHL 001 trial, investigators observed that baseline high tumor burden and inflammatory biomarkers were associated with high CAR T-cell expansion and increased rates of cytokine release syndrome (CRS) and neurotoxicity.

If the model holds up, researchers say they could potentially identify patients at risk for low or high T-cell expansion levels and develop a strategy to enhance or limit the expansion.

TRANSCEND NHL 001 (NCT02631044) is a multicenter, phase 1 trial in relapsed or refractory non-Hodgkin lymphoma evaluating 2 dose levels of JCAR017, also known as lisocabtagene maraleucel, or liso-cel for short.

Liso-cel is a CD19-directed 4-1BB CAR T cell administered at precise doses of CD4+ and CD8+ CAR T cells. It had previously demonstrated high complete remission (CR) rates and low incidences of CRS and neurotoxicity.

Tanya Saddiqi, MD, of City of Hope National Medical Center in Duarte, California, presented data from the dose-finding and expansion cohorts at the 2017 ASH Annual Meeting (abstract 193*).

Study design

Patients with DLBCL after 2 lines of prior therapy or mantle cell lymphoma after 1 prior line of therapy were eligible to enroll in TRANSCEND NHL 001.

Patients with de novo DLBCL, those who transformed from follicular lymphoma, or those with high-grade B-cell lymphoma made up the pivotal or core population. All DLBCL patients enrolled on the trial comprised the full population.

Patients were screened, enrolled, and underwent apheresis. Bridging therapy was permitted while their CAR T cells were being manufactured.

Patients then had a PET scan and lab tests prior to lymphodepletion.

“This is the time point of our interest,” Dr Saddiqi said, “to see if there are any patient characteristics or biomarkers that we can identify . . . that could help us figure out which patients are at higher risk of toxicity, potentially.”

Lymphodepletion consisted of fludarabine (30 mg/m²) and cyclophosphamide (300 mg/m² for 3 days).

Patients received the JCAR017 infusion, and, at specific time points thereafter, cytokine, pharmacokinetic (PK), and clinical lab evaluations were conducted. PK evaluation and scans were performed every 3 months for the first year after JCAR017 infusion, and safety and viral vector follow-up for 15 years.

Dose levels were 5 x 10⁷ cells as a single or double dose (DL1S) and 1 x 10⁸ cells as a single dose (DL2S). Dose level 2 was chosen for further study, and double dosing was discontinued.

“Double dosing was actually not pursued further,” Dr Saddiqi explained, “because it did not seem to add any benefit over single dosing.”

At the time of the presentation, 91 total patients were treated, 67 of whom were the core population.

Results

Dr Saddiqi reported that patients treated with JCAR017 achieved a relatively high best overall response rate (ORR) and high durable CR rates.

“And this seems to be especially true for the core set of patients and particularly for patients at dose level 2,” she added.

At all dose levels, the core patients had a best ORR of 84% (41/49) and a CR rate of 61% (30/49).

At follow-up of 3 months or longer, the core group had an ORR of 65% (26/40) for all dose levels, 52% (11/21) for dose level 1, and 80% (12/15) for dose level 2.

The 3-month CR rate was 53% (21/40) for all dose levels in the core group, 33% (7/21) in dose level 1, and 73% (11/15) in dose level 2.

Dr Saddiqi noted that CRS and neurotoxicity did not differ by dose level or schedule, and there were no grade 5 events of CRS or neurotoxicity.

“Among the core group, dose level change did not add to their toxicity,” she said. “And so the question is: Is it patient factors, is it tumor factors? What is it that is actually causing the toxicities in these patients?”

Dr Saddiqi focused the presentation on patient factors.

Patient factors

The data showed that tumor burden and lactose dehydrogenase (LDH) levels were higher in patients with CRS and neurotoxicity.

Univariate analysis revealed that CRS and neurotoxicity were associated with a shorter time since diagnosis.

However, prior number of therapies, patient weight, and disease stage were not associated with CRS or neurotoxicity.

Investigators were able to identify preliminary risk boundaries. Core patients with high LDH levels (≥ 500 U/L) and sum of the products of diameters (SPD) ≥ 50 cm² at baseline had an 8-fold increase in risk of CRS and neurotoxicity.

“Inversely, if these patients did not meet the cutoff for LDH or SPD,” Dr Saddiqi pointed out, “if they were lower than that, they have significantly lower CRS and neurotoxicity events.”

Investigators also observed that baseline markers of inflammation and inflammatory cytokines trended higher in patients with CRS and neurotoxicity. For CRS, this includes ferritin, C-reactive protein (CRP), IL-10, IL-15, IL-16, TNFα, and MIP-1β. For neurotoxicity, this includes ferritin, CRP, d-Dimer, IL-6, IL-15, TNFα, and MIP-1α.

The team also observed that tumor burden, baseline markers of inflammation, and inflammatory cytokines trended lower in core patients with durable responses.

“Interestingly, it’s inversely true that patients who did have these higher levels [of inflammation markers], and higher tumor burden, and higher LDH, actually were the ones that were either showing no response at 3 months or had lost their response by the 3-month assessment point,” Dr Saddiqi explained.

And in patients with higher baseline tumor burden and inflammatory cytokine levels, JCAR017 T-cell expansion trended higher.

“Some were deemed to be super expanders because their CAR T-cell levels were very high in their blood,” she added.

The investigators created a preliminary logistic model based on the data that suggests a therapeutic window might be able to limit toxicity and optimize efficacy.

The model indicates that patients with higher tumor burden, higher LDH, and higher inflammatory state at baseline seem to be the ones who are having more CRS and more neurotoxicity after CAR T-cell infusion.

“They are expanding their cells much more, yet their responses at 3 months seem to be affected adversely by this entire situation,” Dr Saddiqi said.

"One explanation, potentially, could be that these CAR T cells are seeing a lot of antigen when they go into the body. They have the perfect cytokine milieu to grow, expand, and go crazy in the body, if you will, and very quickly peter out as well because there’s T-cell exhaustion that happens rather rapidly and clinical responses are then then lost.”

The investigators believe that if they can identify those patients ahead of time who may be at risk of too high expansion or too low expansion of their CAR T cells, they may be able to find strategies to push expansion into the “sweet spot of CAR T-cell expansion and ultimately get the holy grail of having durable responses for all with minimal toxicity,” Dr Saddiqi concluded.

TRANSCEND NHL 001 is sponsored by Juno Therapeutics, Inc. Dr Saddiqi has served on a steering committee for JCAR017.

*Data in the presentation differ from the abstract.

Risk stratification may be possible with JCAR017