CAR T-Cell Therapy is Well Tolerated With Promising Efficacy in High-Risk NHL

CAR T-Cell Therapy is Well Tolerated With Promising Efficacy in High-Risk NHL

Dr. Jeremy S. Abramson presents Abstract 120 during Oral Abstract Session B.
A CD19-directed CAR T-cell therapy known as JCAR017 showed potent and durable responses in a study of patients with poor-prognosis relapsed or refractory non-Hodgkin lymphoma (NHL; Abstract 120). The therapy was generally well tolerated, with low rates of severe cytokine release syndrome (CRS) and neurotoxicity.

“Relapsed/refractory B-cell lymphoma is associated with very poor outcomes with modern therapies,” presenter Jeremy S. Abramson, MD, of the Massachusetts General Hospital Cancer Center, said Oral Abstract Session B on CAR T cells on January 26. He noted that few patients achieve durable remissions to available treatments. Aggressive, relapsed/refractory B-cell NHL has an objective response rate of less than 40% and a complete response rate of less than 20% to available therapies.

JCAR017 (lisocabtagene maraleucel, or liso-cel) is a CD19-directed CAR T-cell product comprising individually formulated CD4+ and CD8+ cell suspensions. The patient’s cells undergo immunomagnetic selection, followed by lentivirus transduction, expansion, and formulation at specified composition of CD4+ and CD8+ CAR T cells; these are administered in a precise, flat dose. The CD8+ cells target the tumor, whereas the CD4+ cells do the same in addition to supporting persistence of the treatment.

The TRANSCEND NHL 001 trial was a multicenter study that included both dose-finding and dose-expansion cohorts. Dose level (DL) 1S included 5 x 107 cells in a single dose, DL1D included 5 x 107 cells in a double dose, and DL2S included 1 x 108 cells in a single dose. Patients included in a “core” population had diffuse large B-cell lymphoma or high-grade B-cell lymphoma (de novo or transformed follicular lymphoma) after two lines of therapy; the full cohort also included patients with primary mediastinal B-cell lymphoma and follicular lymphoma grade 3B, or had aggressive lymphoma that had transformed from chronic lymphocytic leukemia or marginal zone lymphoma.

A total of 140 patients underwent leukapheresis, and 108 were treated with JCAR017 following lymphodepletion with fludarabine and cyclophosphamide; six of those were not yet evaluable, and 11 received nonconforming JCAR017, leaving 91 patients evaluable for safety (45 at DL1S, six at DL1D, and 40 at DL2S). Of those, 67 patients were considered the core population. The full cohort had a median age of 61, and most patients (89%) had an ECOG performance status of 0 or 1. Patients had a median of three prior lines of therapy, and 67% were chemorefractory.

Most patients experienced at least one treatment-emergent adverse event (TEAE), and about 60% of patients experienced a TEAE deemed related to JCAR017. The most common TEAE was neutropenia, followed by thrombocytopenia and anemia, and Dr. Abramson said these reflected the cytoreducing chemotherapy.

In the full cohort, 35% of patients had CRS, although almost all cases were grade 1/2. In the core population, 41% of those receiving the DL1S dose had any-grade CRS, compared with 24% at the DL2S dose level. Again, nearly all cases of CRS were grade 1/2.

Neurotoxicity occurred in 19% of the full cohort, and 12% had grade 3/4 neurotoxicity. In the core group, 24% of DL1S and 17% of DL2S patients had any-grade neurotoxicity, and 21% and 7% were grade 3/4, respectively. When considered together, 40% of the full cohort had CRS or neurotoxicity, and 12% had serious cases of either one. In the core cohort, 44% of DL1S and 34% of DL2S patients had either of the two toxicities, and 21% and 7%, respectively, were serious.

Dr. Abramson noted that low rates of toxicity and late onset of those toxicities support outpatient administration of this therapy. The median time to onset of CRS in the full cohort was 5 days, and for neurotoxicity it was 10 days. Hospitalization days per patient were lower when the therapy was given as outpatient, with a mean of 15.6 days for inpatient and 9.3 days for outpatient. When given as an outpatient, the median day of hospitalization was day 5. None of the eight patients who were treated as outpatients required ICU admission.

In the full cohort, the overall response rate (ORR) was 74%, and 52% had a complete response (CR). In the core group, the ORR was 76% with DL1S and 81% with DL2S; there were more CRs with DL2S, at 63% compared with 47%. Half of the DL2S patients with 6 months of follow-up and who were in CR remain in CR.

“Responses are seen across all high-risk subgroups,” Dr. Abramson said. For example, the ORR in patients with transformed follicular lymphoma was 76.9%. In patients with stage III/IV disease, the ORR was 62.2%, and patients who were chemorefractory had an ORR of 64.9%. In patients with double/triple hit lymphoma, the response rate was 84.6%.

Dr. Stephen Gottschalk discusses Abstract 120 during Oral Abstract Session B.
Dr. Abramson said the early overall survival (OS) data “compare favorably” in this high-risk population to what is seen with available therapies. The median OS was not reached in the full cohort, and the 6-month OS was 86%. In those who achieved a CR, the median was not reached, and the 6-month OS was 94%. In those with a partial response, the median was 9 months, and the 6-month OS was 81%. In those who were nonresponders, the median was 6.2 months, and the 6-month OS was 66%.

Dr. Abramson said the “potent and durable remissions” in this high-risk malignancy, along with manageable safety profile, are promising, and a subsequent pivotal cohort using the DL2S dose is now enrolling patients. That cohort will include only patients with diffuse large B-cell lymphoma.

Stephen Gottschalk, MD, of St. Jude Children’s Research Hospital, was the discussant for the session, and he said that some outstanding questions remain regarding the use of CD19-targeted CAR T-cell therapy in NHL. “How durable are these responses? What is the mechanism of failure?” he asked. If patients do achieve a long-term complete response, when is it safe to deplete CD19 CAR T cells to allow the recovery of normal B cells?

Correlates of Response and Toxicity

A second analysis based on the same study looked at the patient characteristics and biomarkers that correlate with clinical outcomes (Abstract 122). Tanya Siddiqi, MD, of the City of Hope National Medical Center, presented results of that analysis during the same Oral Abstract Session on January 26.

“The precise doses that are given to patients make it possible to analyze other data in a better way, because […] there may not be a lot of product variability,” Dr. Siddiqi said.

Patients in the core cohort with a higher tumor burden, measured by the sum of the products of diameters (SPD), at baseline, were more likely to develop CRS during the study (p < 0.001). The same was true for lactate dehydrogenase (LDH), as patients with higher levels at baseline were more likely to develop the complication (p < 0.001).

Similarly, patients with higher SPD at baseline were more likely to develop neurotoxicity over the course of the trial (p = 0.006), and higher baseline LDH levels were also more likely to develop this complication (p = 0.018).

Dr. Tanya Siddiqi presents Abstract 122 during Oral Abstract Session B.
When divided into low and high LDH levels and low and high SPD, the correlation with toxicities was clear. Patients with LDH of at least 500 U/L had approximately an 8-fold increased risk of developing CRS and only a slightly lower risk of developing neurotoxicity. For SPD of at least 50 cm2, the odds ratios were similar. Patients with an LDH less than 500 U/L and an SPD less than 50 cm2 had far lower rates of CRS and neurotoxicity, with odds ratios below 0.25.

Higher JCAR017 CAR T-cell expansion is associated with dose level, with higher Cmax of both CD4+ and CD8+ cells at the DL2 dose (p < 0.05).

The median Cmax for CD4+ was 7.0 in the DL1S group and 14.9 in the DL2S group. The median area under the curve (AUC0–28) was also higher in the DL2S group. Similarly, the median Cmax for CD8+ was 62.8 with DL2S and 26.1 for DL1S, and the median AUC0–28 was also higher with DL2S.

Higher baseline tumor burden also appeared to be correlated with higher JCAR017 CAR T-cell expansion, with a correlation coefficient of 0.22 (p = 0.010). The same was true for higher levels of baseline inflammatory cytokine levels, with a significant correlation between cell expansion and interleukin (IL)-15, macrophage inflammatory protein-1α, and tumor necrosis factor (TNF)-α (p < 0.05; correlation was not significant for IL-7).

Baseline markers of inflammation and inflammatory cytokine levels also trend higher in patients who develop CRS and neurotoxicity.

Interestingly, those same baseline markers of inflammation, levels of inflammatory cytokines, and tumor burden all may trend lower in patients with a durable response to JCAR017. Lower levels of a variety of markers were seen in patients who had a response at 3 months, including LDH, ferritin, IL-6, IL-10, TNF-α, CXCL10, and others (p < 0.05 for all); the association was not significant for SPD, however.

“The theory is that these patients with a higher tumor burden and a higher inflammatory state have such a rapid expansion […] that those T cells may be getting exhausted or dying out very quickly, and patients lose their response,” Dr. Siddiqi said.

She added that preliminary logistic modeling suggests there is a therapeutic window of CAR T-cell expansion that could limit toxicity and optimize the therapy’s efficacy. “If you look at patients who had lower T-cell expansion for whatever reason, those patients may have lower toxicities, but they also have lower overall response rates and certainly lower durability of that response at 3 months,” she said. “Alternatively, if patients have super high expansion of CAR T cells, they have higher toxicity and higher response rates, but they lose those responses quickly as well.” If strategies could be devised that nudge patients into that window of T-cell expansion, both toxicity and response might be optimized.

Malcolm K. Brenner, MD, PhD, of Baylor College of Medicine Texas Children’s Hospital, was the discussant for the abstract. He said that such a refinement of patients might be difficult, and that it may be more likely that adding in some sort of control mechanism to the therapy might be necessary. There are possibilities for such mechanisms that work via apoptotic, cytolytic, or antiproliferative mechanisms, among others.

–Dave Levitan