γ-Secretase inhibitor in combination with BCMA chimeric antigen receptor T-cell immunotherapy for individuals with relapsed or refractory multiple myeloma: a phase 1, first-in-human trial.

Abstract

Background

Gamma secretase inhibitors (GSIs) increase B cell maturation antigen (BCMA) density on malignant plasma cells and enhance anti-tumor activity of BCMA CAR T cells in preclinical models. We report results from a clinical trial of 18 patients with relapsed/refractory multiple myeloma (MM) combining escalating doses of a BCMA-targeted therapy with a GSI.

Methods

Eligible patients were ≥ 18 years old, had relapsed/refractory MM, prior autologous stem cell transplant or persistent disease after > 4 cycles of induction therapy, ECOG ≤ 2, and could have received prior BCMA targeted therapy. The primary endpoints were to evaluate the safety and recommended phase 2 dose (RP2D) of BCMA CAR T cells in combination with crenigacestat, an oral GSI. To assess the impact of the GSI on bone marrow plasma cell BCMA surface density, patients received GSI during a pre-treatment “run-in” consisting of three doses administered 48 hours apart. BCMA CAR T cells were infused at a starting dose of 50 × 10⁶ CAR+ cells, in combination with the GSI dosed three times weekly for up to 9 doses. The trial is registered with ClinicalTrials.gov, NCT03502577, and has met accrual goals.

Findings

Eighteen patients with MM were treated from July 11, 2018 to April 14, 2021, with a median follow up of 36 months (95% CI, 26 months to not reached). Among the 10 females (56%) and 8 males (44%), 7 had prior treatment with a BCMA targeted agent. Three oral doses of crenigacestat during the “run-in” resulted in a median 12.2-times increase in BCMA density, (range, 3.18 – 156.58, Interquartile range ([IQR] 9.83 to 30.51-times). The most common non-hematologic AEs ≥ grade 3 were hypophosphatemia in 14 (78%), fatigue in 11(61%), hypocalcemia in 9 (50%), and hypertension in 7 (39%). Two deaths reported outside of the 28-day adverse event collection window were related to treatment. Patients were treated at doses up to BCMA CAR 450 × 10⁶ CAR+ cells and the RP2D was not reached.

Interpretation

Combining a GSI with BCMA CAR T cells is safe, and crenigacestat increases target antigen density. Deep responses were observed amongst heavily pre-treated BCMA exposed and naïve MM patients. Further study of GSIs given with BCMA targeted therapeutics is warranted in clinical trials. This trial was funded by Juno Therapeutics, a Bristol Myers Squibb Company.

Introduction

Proteasome inhibitors (PI), immunomodulatory agents (IMiD), CD38 antibodies, and autologous stem cell transplantation (ASCT) have improved survival for multiple myeloma (MM) patients(1, 2). However, almost all patients ultimately relapse. Early efforts targeting BCMA (B-cell maturation antigen; also known as TNFRSF17) with CAR T cells in MM proved successful in treating patients after multiple lines of prior therapy(3–5) and two agents, idecabtagene vicleucel and ciltacabtagene autoleucel have been approved by the FDA.

Despite promising results from early studies of BCMA CAR T therapy for relapsed refractory MM, curative potential has yet to be demonstrated. The mechanism of post-treatment relapse is of interest and putative contributory factors include: inadequate persistence of the CAR T cells, antigen loss/escape, and an inhibitory immune microenvironment(6). BCMA is largely restricted to plasma cells, and is cleaved from the cell surface via the enzymatic activity of cell membrane gamma secretase(7). Reduced antigen density and increasing soluble BCMA (sBCMA) levels can adversely affect anti-tumor efficacy of BCMA targeted therapies(7). Gamma secretase inhibitors (GSIs) abrogate enzymatic function and increase BCMA surface density on plasma cells in vivo with a concomitant decrease in sBCMA levels(7, 8). GSIs can enhance target recognition and thus the anti-MM efficacy of anti-BCMA CAR T cells in preclinical systems(8).

We report the results of a phase 1, first-in-human clinical trial combining an oral GSI, crenigacestat, with a BCMA CAR T-cell product.

Methods

Study Design and Participants

This phase 1 Institutional Review Board (IRB) approved trial was conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization guidelines for Good Clinical Practice as a single center study at Fred Hutchinson Cancer Center (FHCC) in Seattle, WA, USA.

Eligible patients included those with relapsed or treatment refractory MM ≥ 18 years of age, ECOG performance status of 0–2, life expectancy > 3 months, with measurable disease by International Myeloma Working Group (IMWG) criteria, and ≥ 10% CD138+ malignant plasma cells by immunohistochemistry on bone marrow (BM) core biopsy (appendix pp 3–4) (9). Sex and ethnicity were self reported. Patients had to have either undergone ASCT or be transplant ineligible with disease persisting after >4 cycles of induction therapy and refractory after both a PI and IMiD; demonstrate adequate organ function (estimated renal function with creatinine clearance ≥ 20 ml/min); and have adequate hematologic function (ANC ≥ 1000/mm³, Hb ≥ 8 g/dl, and platelet count > 50,000/mm³, unless cytopenias were determined related to underlying MM). Laboratory assessment of serum M protein, immunoglobulins, serum free light chains, BM biopsy and aspirate, and imaging for osseous lesions was required for enrollment. Patients previously exposed to BCMA targeted therapy, including CAR T therapy, were permitted to enroll. Patients could be removed from the trial by self-choice or by investigator decision.

Exclusions included documented malabsorptive syndromes including enteropathies, gastroenteritis (acute or chronic), or diarrhea (acute or chronic); known AL amyloidosis; active auto-immune disease; or > 1 hospital admission lasting 5 days or more for documented infection in the prior 6 months. Further details regarding eligibility criteria can be found in the appendix pages 3–4.

Procedures

The product, FCARH143, was manufactured by the Fred Hutchinson Therapeutic Products Program from autologous T cells collected via leukapheresis. CD4 and CD8 subsets were selected using immunomagnetic selection (CliniMACS, Miltenyi Biotec, Bergisch Gladbach, Germany) transduced separately using an FCARH143 lentiviral vector containing the anti-BCMA CAR, and subsequently expanded(10). Chimeric antigen receptor T cells were then formulated in a 1:1 ratio of CD4:CD8 CAR T cells (+/− 15%) and cryopreserved for bedside thaw and administration.

Prior to any bridging therapy or lymphodepleting chemotherapy, all patients received 3 doses of the GSI, crenigacestat, at 25 mg by mouth every other day, and a BM biopsy was obtained within 12 hours of the third dose of crenigacestat during this “run in” phase for assessment of plasma cell surface BCMA density as compared to level at screening.

Bridging treatment prior to lymphodepleting (LD) chemotherapy and CAR T cells was permitted; adequate renal function, defined as eGFR ≥ 20 ml/min, and the absence of uncontrolled active infection were required.

LD chemotherapy consisted of cyclophosphamide, 300 mg/m² intravenous (i.v.); days 1–3, and fludarabine, 25 mg/m² (i.v.) days 1–3. BCMA CAR T cells were administered within 72 hours following LD in 16/18 patients. Doses of 50 × 10⁶, 150 × 10⁶, 300 × 10⁶, and 450 × 10⁶ total CAR+ T cells were examined using a modified toxicity probability interval (mTPI-2) algorithm, with a targeted DLT rate of 30%(11). Crenigacestat was scheduled as 25 mg three times weekly starting on day of CAR T infusion, for a total of 9 doses.

Laboratory assessments of hematologic function, chemistry, serum M protein, and serum free light chains, were performed at screening, prior to lymphodepletion, on days +14, +28, +60, +90, +180, and +365. Imaging assessments for osteolytic disease were performed at screening, prior to lymphodepletion, day +28, and day +90. All ≥ grade 3 adverse events were collected during leukapheresis, for 96 hours post GSI run-in and from of the start of T cell infusion through day 28. All adverse events were grade by the Common Terminology Criteria for Adverse Events criteria, version 4.0 (12). For monitoring of potential risks and toxicity related to CAR T cells, mini-mental status examination-2 was performed within 72 hours of infusion, and vital signs were evaluated prior to CAR T cell infusion, every 15 minutes during, and then hourly for 2 hours after the infusion; C-reactive protein, serum ferritin, complete blood count, and serum chemistries were obtained prior to infusion and then on days 1, 3, 7, 10, 14, 21, 28, and as clinically indicated. For GSI-associated toxicities, holding doses of GSI was at discretion of lead investigator.

Outcomes

The primary outcomes of the study included safety as measured by NCI CTCAE v 4.0 (12), and identification of the RP2D. Safety was also assessed by clinical evaluation (including history and physical, pulse oximetry, CBC and chemistry, before and at intervals after the T cell infusion), serum cytokine levels, B cell reconstitution, serum immunoglobulin levels, replication competent lentivirus, and adverse event reporting. Relevant safety findings are included in the results section and appendix. The secondary outcomes included determination of the peak concentration, in vivo persistence and phenotype of CAR T cells (as measured by qPCR and flow cytometry), and estimation of the antitumor activity of the combination of BCMA CAR T cells with the GSI (as defined by ORR determined by the IMWG response criteria(13) [including rates of stringent complete response, complete response, very good partial response, and partial response], PFS and OS). Exploratory outcomes were determination of the effect of the GSI on surface BCMA expression levels on malignant plasma cells, determination of the effect of the GSI on sBCMA levels, and determination of the degree to which CAR T cells traffic to the BM or other tumor sites, establish memory cell populations, and function in-vivo. A comprehensive evaluation of the secondary outcome related to CAR T cell trafficking and function is ongoing as patients are being monitored for relapse. This secondary outcome will be reported separately once a complete dataset is available.

CRS was determined by the Lee criteria and neurologic toxicity was assessed the NCI CTCAE symptom scoring(14, 15) and ICANS ASTCT scoring system (retroactively applied). In any case where grading was not easily discerned, a detailed chart review was performed, and scoring adjudicated by 2 physicians experienced in assessment of ICANS. Responses were assessed by IMWG criteria (13).

Statistical Analyses

Demographics and relevant patient features were summarized using descriptive statistics, such as median and Interquartile range (IQR), and range for continuous variables, and counts and percentages for categorical variables (Table 1). A p-value of < 0.05 was considered significant. Kaplan-Meier curves were used to estimate the OS and PFS, and log-rank tests were conducted. Duration of response (DOR) was conducted as a post-hoc analysis, as was comparison of efficacy amongst BCMA naïve to BCMA treated populations. The median time to event and 95% confidence intervals were provided. The cutoff for the survival analysis was August 23, 2022. For time to event outcomes, such as PFS and OS, post hoc univariate analysis was conducted using Cox regression to assess the association between a risk factor and each outcome of interest. The proportional hazards assumption was checked by Schoenfeld residuals. Post hoc landmark analyses were conducted to evaluate the effects of day 60 sBCMA, and subjects with events, or censored before day 60, were removed accordingly. Three patients were excluded due to death before day 60 and 2 due to insufficient sBCMA data. Post-hoc analysis of time to platelet recovery after CAR T infusion was performed. All statistical analyses were conducted using R version 4.2.1(16).

Table 1.

Demographics and Patient Characteristics

	Total (n=18)
Median age, years	65 (Range, 44–75, IQR 56.5 – 69.8)
Sex
Male	8 (44%)
Female	10 (56%)
Ethnicity/Race
Black/African American	1 (6%)
White/Caucasian	17 (94%)
Median time from diagnosis	6.2 years (Range, 2–13, IQR 2.8 – 7.6)
Median number of prior regimens	10 (Range, 4–19, IQR, 7–12.5)
Bortezomib refractory	15 (83%)
Carfilzomib refractory	15 (83%)
Lenalidomide refractory	18 (100%)
Pomalidomide refractory	16 (89%)
Daratumumab refractory	16 (89%)
PI/IMID Refractory	18 (100%)
PI/IMID/CD38 Refractory	16 (89%)
Prior CNS involvement	1 (6%)
Extramedullary disease	5 (28%)
≥ 1 High risk CA	13 (72%)
≥ 1 High risk CA (17p-; t(4;14); t(14;16) only	10 (55%)
≥ 2 High risk CA	5 (28%)
Receipt of bridging therapy	12 (67%)
Prior BCMA Directed Therapy	7 (39%)
BCMA ADC	2 (11%)
BCMA CAR T	4 (22%)
BCMA T cell engager	1 (6%)
Prior autologous stem cell transplant	16 (89%)
Prior allogeneic stem cell transplant	2 (11%)

Dose escalation at 4 different dose levels, with a starting dose of 50 × 10⁶ CAR+ cells, was conducted using a mTPI-2 algorithm(11). The mTPI-2 approach was followed until a total of 18 patients evaluable for DLT were treated; the recommended dose following the 18^th patient was to be regarded as the RP2D. This study is registered with ClinicalTrials.gov, number NCT03502577.

Role of the Funding Source

The clinical trial funding source reviewed the final manuscript but did not design the trial, collect, analyze, or interpret data. The corresponding author had full access to all the data and the final responsibility to submit for publication.

Results

From July 11, 2018 to April 14, 2021, 19 patients underwent leukapheresis and run in with crenigacestat. CAR T product manufacturing was successful for all patients. Two (16%) of the 12 patients who received bridging therapy responded (both partial responses; appendix pp 5). One patient (5%) of 19 did not proceed with BCMA CAR T infusion due to decline in performance status related to disease progression. The CAR T cell product was comprised of the planned CD4:CD8 ratio of 1:1 (+/− 15%) in 17/18 patients. One patient required repeat leukapheresis for sterility/quality control. Of the 18 patients, 5 (28%) received 50 × 10⁶ CAR+ cells, 3 3 (17%) received 150 × 10⁶ CAR+ cells, 3 (17%) received 300 × 10⁶ CAR+ cells, and 7 (39%) received 450 × 10⁶ CAR+ cells.

Patient demographics are shown in Table 1. Prior to lymphodepleting chemotherapy, the median BM plasma cell involvement was 45% (range, 10 – 90%, IQR 20%−80%) and 8 (44%) of 18 patients had BM involvement with ≥ 70% plasma cells by CD138 immunohistochemistry. Six patients (33%) of 18 patients had severe or very severe thrombocytopenia(17) prior to lymphodepletion including counts that were < 25,000/μL (4 patients [22%] of 18), and < 50,000/μL (2 patients [11%] of 18). Seven patients (39%) of 18 previously received BCMA targeted therapy (Table 1). The best responses to prior BCMA therapy were PD in 3, SD in 2, and VGPR in 2.

Adverse events (AEs), NCI CTCAE V 4.0 ≥ grade 3 are depicted in Table 2. The median time to platelet recovery ≥ 50,000/μL post CAR T infusion was 38 days (IQR, 24–64 days), and 66% of patients had recovery to ≥ 50,000/μL 60 days post CAR T infusion. Hospitalization was necessary for 17 (94%) of 18 subjects; the median hospital duration was 11 days (range, 3–48 days, IQR 8 – 16 days). Grade 3 or higher infections occurred in 3 (16%) of 18 patients (Table 2).

Table 2.

Adverse Events, NCI CTCAE Grade 3 or higher, and IEC-related toxicities of interest

Adverse Event*	Grade 3		Grade 4
	n	%	n	%
Any	18	100%	17	94%
Gastrointestinal
Anorexia	6	33%	0	0%
Diarrhea	3	17%	0	0%
Mucositis	2	11%	0	0%
Nausea	2	11%	0	0%
Upper GI Hemorrhage	0	0%	1	6%
Colitis	1	6%	0	0%
Non-Hematologic
ALT increased	3	17%	1	6%
AST increased	3	17%	2	11%
Fatigue	11	61%	0	0%
Hypertension	6	33%	1	6%
Hypocalcemia	7	39%	2	11%
Hypokalemia	3	17%	0	0%
Hyponatremia	4	22%	0	0%
Hypophosphatemia	12	67%	2	11%
Hypotension	5	27%	1	6%
Pneumonia	2	11%	1	6%
Hypoxia	7	39%	1	6%
Febrile Neutropenia	15	83%	0	0%
Sinus tachycardia	3	17%	0	0%
Infections	2	(11%)	2	(11%)
Hematologic
Anemia	17	94%	0	0%
Leukopenia	3	17%	15	83%
Thrombocytopenia	3	17%	14	78%
Neutropenia	0	0%	15	83%
Lymphopenia	1	6%	17	94%
IEC-Related Toxicities of Interest	Any grade		Grade 3 and 4
Cytokine release syndrome (Lee)	17 (94%)		5 (28%)
Cytokine release syndrome (ASTCT)	17 (94%)		2 (11%)
Immune effector cell mediated neurologic syndrome (ICANS)	7 (39%)		2 (11%)

^*Shown here are grade 3 or higher adverse events, not including symptoms of cytokine release syndrome or immune effector cell associated neurologic syndrome, occurring in 11% or more of patients from the start of T cell infusion through day 28.

Serious adverse events were documented in 17 (94%) of 18 participants (appendix p 6). Two treatment-related deaths occurred, outside of the 28-day predetermined reporting period for adverse events. One participant had a treatment-related death on day 34 due to disseminated drug-resistant herpes simplex virus. One treatment-related death occurred on day 35. This participant had an Eastern Cooperative Oncology Group score of 3 with several comorbidities, and inclusion was subsequently limited to participants with a score of less than 3. This participant had a dose-limiting toxicity, grade 4 cytokine release syndrome and grade 4 immune effector cell-associated neurotoxicity syndrome, in the setting of disseminated Aspergillus infection with involvement of brain parenchyma identified postmortem. Per dose escalation design, an additional two participants were treated a dose of 50 × 106 CAR+ cells. One participant had no increase in BCMA density after the GSI run-in (participant 11, figure 2). This individual (treated previously with a BCMA antibody-drug conjugate) had baseline thrombocytopenia before treatment (ie, platelet count of 27 000 per μL), had no response to the BCMA CAR T-cell therapy, and died from complications of a fall and traumatic brain haemorrhage on day 33. In total eight participants have died; two deaths were treatment-related (as already described), four deaths were with known multiple myeloma progression, and two deaths were due to other causes (ie, one due to traumatic haemorrhage as already described and one due to undetermined causes). Although the recommended phase 2 dose was not reached, we recommend the dose for phase 2 evaluation to be 450 × 106 BCMA CAR+ T cells in combination with crenigacestat.

Figure 2.

Kaplan Meier plot for PFS (A) and OS (B)

Among the 17 (94%) of 18 treated patients who had an increase in BCMA antibody binding capacity (ABC), the BCMA ABC increased from a median of 663 to 9583 receptors per cells, or a median of 12.2-times (range, 3.2-times to 156.6-times, IQR 9.83 to 30.51-times) (Figure 1). Among patients previously exposed to BCMA targeted therapy, the absolute increase in BCMA surface density was significantly lower than the increase observed in BCMA treatment naïve patients (p=0.0019) [appendix pp 10]. Amongst the 17 patients who had an increase in BCMA with the GSI, sBMCA decreased by a median of 317 ng/mL (IQR −704 to −117, range, 28.8 to 2771 ng/mL) over the 5-day run-in (appendix p 9). The run-in GSI administration was well tolerated.

Figure 1.

Pre- and post- GSI run-in BCMA density as measured by antibody binding capacity (ABC) on BM plasma cells obtained from patients before, and after, three oral doses of GSI. The ABC is measured by flow cytometry and represents the number of potential binding sites for an anti-BCMA antibody. The ABC increased in all but 1 patient. Patients are separated by a history of prior BCMA targeted therapy before enrollment (Yes/No).

Responses are depicted in Table 3. The median duration of response (DOR) was 14.4 months (95% CI, 5.9 to not reached). The only 2 patients (11%) of 18 who did not respond had stable disease; both patients had previously received BCMA targeted therapy. The median time to best response was 2.5 months (range, 1 to 12 months, IQR 1 to 12 months). Notably, one patient (50 × 10⁶ CAR+ cells dose level) has remained in a sCR for over 4 years. Responses improved over time; amongst the patients with VGPR or better, 8 of 14 patients continued to have a detectable serum monoclonal protein until 12 months post-CAR infusion (appendix p 12).

Table 3.

Efficacy of BCMA CAR T cells combined with a GSI

	Patients (n=18)	BCMA Naïve (n=11)	BMCA Exposed (n=7)
Best overall response	89% (16)	100% (11)	71% (5)
Stringent CR	8 (44%)	8 (73%)	(0)
Complete response	11% (2)	18% (2)	0% (0)
Very good partial response	4 (22%)	1 (9%)	3 (43%)
Partial response	11% (2)	0% (0)	29% (2)
Stable disease	11% (2)	0% (0)	29% (2)
Progressive disease	0% (0)	0% (0)	0% (0)
Survival (95% CI)
Progression free survival	11 months (5.4 – not reached)	28.8 months (19.2-not reached)_	2.6 months (1.1-not reached)
Overall survival	42 months (13.2 – not reached)	42 months (42-not reached)	6.8 months (1.1-not reached)

With a median follow-up of 36 months (95% CI, 26 to not reached), PFS for all patients was 11 months (95% CI, 5.4 to not reached) (Figure 2; Table 3). The median OS was 42 months (95% CI, 13.2 months to not reached) (Figure 2; Table 3).

Fifty percent of the treated patients received all pre-planned doses of crenigacestat; 5 (28%) received 8/9 of planned doses, 2 (11%) received 7/9 of planned doses, 1 (5%) received 6/9 of planned doses, and 1 (5%) received 4/9 of planned doses. GSI doses were held due to CRS/ICANS (n=3), mucositis (n=1), patient declining dose (n=1), nausea/vomiting (n=1), and unspecified (n=3).

Any grade CRS was observed in 94% of patients (Table 3). By the Lee 2014, Grade 1 CRS occurred in 4 patients (22%), Grade 2 CRS in 8 patients (44%), Grade 3 CRS in 4 patients (22%), and Grade 4 CRS in 1 patient (6%). When applying the ASTCT consensus criteria retroactively, Grade 1 CRS occurred in 9 patients (50%), grade 2 CRS in 6 patients (33%), grade 3 CRS in 1 patient (6%), and grade 4 CRS in 1 patient (6%). Eleven patients (61%) received tocilizumab for treatment of CRS, 1 dose (n=6) and 2 doses (n=5). Dexamethasone was administered in 12/18 patients (66%), with a median cumulative dose of 20 mg (range 10 mg to 640 mg, IQR 20 mg to 182.5 mg) and 4/18 patients required ICU admission for CRS. Median time to onset of CRS was 1 day (range, 0 – 11 days, IQR 1– 4 days). Median duration of CRS was 4 days (range, 1 – 34 days, IQR 1 – 5 days). There was no correlation between CRS grade and BCMA ABC on day 5 of run in (appendix p 11). There was also no correlation between peak CAR T expansion and development of CRS or ICANS.

All neurologic changes by CTCAE v 4.0 were documented. ICANS was retrospectively graded using ASTCT criteria(15). ICANS (any grade) was present in 7 patients (39% including 2 grade 1, 3 grade 2, 1 grade 3, and 1 grade 4 [Table 3]). Median time to onset of ICANS was 4 days (range, 0 – 19 days, IQR 0.8 – 9.5 days), and median duration of ICANS was 7 days (range, 1 – 24 days, IQR 2.5 – 8 days). ICANS fully resolved in all but one patient who declined early treatment with tocilizumab and dexamethasone. All 7 patients received steroid therapy and 2 patients received the interleukin 1 receptor antagonist anakinra. The BCMA ABC value on day 5 of the run in was associated with the risk of developing ICANS (p=0.0059) (appendix p 11).

Analysis of CAR T cell persistence by quantitative polymerase chain reaction (qPCR; Qiagen, Germantown, USA) was based on a data cut off December 21, 2021. Median time to the maximum CAR T cell expansion (T_max) was 15 days (range, 7–98 days, IQR 13–21 days) (appendix p 17). C_max was significantly higher in the 450 × 10⁶ CAR+ cell dose level (p = 0.042, appendix p 17) when compared to the dose level 50 ×10⁶. Flow cytometry (LSRII or Symphony 4; BD Biosciences, San Jose, USA) measuring BCMA+/truncated EGFR+/CD3+ CAR T cells in the peripheral blood confirmed the qPCR finding (appendix p 18).

Receipt of prior BCMA targeted therapy appeared to impact outcomes (Table 3, Figure 3). ). Amongst BCMA naïve patients, the median duration of response was 791 days. Amongst patients without prior exposure to BCMA targeted therapy (n = 11), the median PFS was 28.8 months (95% CI, 19.2 months to not reached), while for those exposed to prior BCMA targeted therapy (n = 7), the median PFS was only 2.6 months (95% CI, 1.1 month to not reached) (Figure 3). The median OS for patients without prior BCMA therapy was 42 months (95% CI, 42 months to not reached), while for those with prior BCMA therapy, it was 6.8 months (95% CI, 1.1 month to not reached) (Figure 3). Neither the presence nor number of high-risk chromosomal changes correlated with PFS or OS.

Figure 3.

Kaplan Meier plot for PFS (A) and OS (B), by receipt of prior BCMA Targeted therapy

For all patients, univariate analysis revealed that BCMA ABC on day 5 of the GSI run in was associated with improvement in PFS (HR = 0.094, p = 0.006) and OS (HR = 0.109, p = 0.017) (appendix pp 8, 14). Day 60 sBCMA was significantly associated with PFS (HR= 24.08, 95% CI 3.07–189.06, p = 0.002) and OS (HR=11.71, 95% CI, 1.69–81.03, p = 0.013) in a univariate Cox proportional hazards model (appendix pp 8, 15).

CAR T cell expansion kinetics did not impact outcomes (appendix p 8). Higher CAR T exposure, as measured by the area under the curve of the PCR level from time of first dose to 28 days (AUC₂₈), was not associated with PFS or OS in univariate analysis (p = 0.4 and 0.057, respectively). The peak CAR T expansion (C_max)was not significantly associated with PFS or OS (p = 0.96 and 0.18, respectively by univariate analysis). There was no association between prior receipt of BCMA targeted therapy and peak CAR T cell expansion (p=0.66).

Amongst subjects with disease progression, 80% (8/10) had no evidence of persisting CAR T cells by qPCR. One of the two patients relapsing in the setting of persistent measurable CAR T cells underwent BM biopsy (see appendix page 20).

Discussion

This first-in-human phase 1 trial of a BCMA directed agent in concert with a GSI (crenigacestat) demonstrates that the combination is both feasible and safe in patients with relapsed/refractory MM. Crenigacestat led to striking increases in BCMA surface density on malignant plasma cells, decreased sBCMA levels, and deep responses in heavily pre-treated high-risk MM patients. The study population included individuals otherwise ineligible for BCMA CAR T cells on multicenter clinical trials (when applying the enrollment criteria for the CARTITUDE-1 and KarMMA studies, 50% of patients reported here would have been ineligible to enroll on those trials)(18, 19). A comprehensive assessment of efficacy is beyond the scope of this phase 1 dose escalation trial; however, we note that among patients not previously exposed to a BCMA targeting agent, the PFS was 28 months and OS 42 months, despite 36% of treated patients receiving the starting dose level (5 × 10⁶ CAR+ T-cells). The first subject treated at the 50 × 10⁶ CAR+ cell dose level, achieved a stringent CR and remains disease free over 4 years after treatment. This patient had a BCMA ABC that increased 33-times during the 5-day run in (406 [baseline]; 13349 [day 5]). Despite the dose escalation design, the CR/sCR rate for BCMA treatment naïve patients receiving FCAR H143 with GSI was 100%. As comparison, the updated CR/sCR rate for patients receiving the FDA approved dose of ciltacabtagene autoleucel on CARTITUDE-1 was 83%(20) and the initial rate was 67%(18). Individuals on CARTITUDE −1 received a similar number of prior regimens (median = 6) to BCMA naïve patients receiving FCAR H143 with GSI on our trial (median =7). Other differences between the CARTITUDE −1 and FCAR H143 patient populations suggest that comparing responses between ciltacabtagene autoleucel and FCAR H143 with GSI may underestimate the benefit from the GSI combination. FCAR H143 recipients harbored more treatment resistant disease (18% response rate to bridging therapy for FCAR H143 versus 45% response rate to bridging before ciltacabtagene autoleucel). When comparing the same “high-risk” cytogenetic features, 24% of patients on CARTITUDE-1 harbored 17p-, t(4;14) and/or t(14;16), while one or more of these features was present in 73% (n= 8/11) of the BCMA naïve patients who received FCAR H143 with GSI.

The dose and duration interval of GSI treatment were selected in consideration of two factors, 1) clinical safety/tolerance data from prior patient trials with crenigacestat and, 2) avoidance of potential CAR T cell impairment through GSI mediated inhibition of Notch signaling at higher doses. Crenigacestat was previously administered to patients with advanced or metastatic cancer in an open-label phase 1 dose escalation clinical trial, in which Notch pathway inhibition was a desired outcome. Pharmacokinetic data revealed that the RP2D dose (50 mg TIW) resulted in >50% inhibition of Notch-regulated genes, with 45 mg TIW reported to be the minimal biologic efficacious dose to achieve Notch inhibition(18). Our group previously postulated that crenigacestat at lower concentrations than studied previously did not affect T-cell division after tumor cell stimulation(8). To reduce the theoretical risk of diminished CAR T cell function, we selected a dose of the GSI for BCMA CAR T cell combination that was previously proven safe, well tolerated and substantially lower than the dose previously found to inhibit Notch.

The primary aim of this phase 1 trial was to evaluate safety, toxicity, and identify the RP2D of BCMA CAR T cells for combination with crenigacestat. Our efficacy findings suggest a potential role of GSI in augmenting outcomes with BCMA CAR T cell therapy, however the lack of a comparator arm limits interpretation. The optimal timing and frequency of GSI administration after BCMA CAR T cell administration was not examined in our trial, which is another limitation.

All grade CRS was observed in 94%, however 50% were grade 1, 39% were grade 2, and symptoms resolved at a median of 4 days. ICANS was noted in 38%; however, symptoms fully resolved at a median of 7 days for all patients, excluding one individual who initially declined treatment for CRS/neurological changes and whose assessments were complicated by a disseminated fungal infection involving brain parenchyma on autopsy. Gastrointestinal side effects and thrombocytopenia, both expected toxicities with a gamma secretase inhibitor, were present, but were self-limited.

All patients with available BM tumor specimens at disease progression demonstrated a significant reduction in BCMA ABC when compared to GSI day 5 run-in values. In the one patient with a BM biopsy revealing progression in the setting of persisting CAR T cells, the BCMA ABC was 148-times reduced. These findings support the hypothesis that BCMA antigen low MM cells may predominate in the early stages of progression after BCMA CAR T cell therapy(4) and this low antigen density may contribute to disease progression(19).

We found that both the change in MM cell BCMA surface density following 3 GSI doses, and the absolute BCMA ABC after the third dose, were significant biomarkers associated with PFS and OS. Previous reports have not suggested association between response and BCMA quantitation (20–22). In contrast to our trial, which examined BCMA density continuously by flow cytometry, most prior BCMA trials appear to have retrospectively identified thresholds for BCMA expression using immunohistochemistry (e.g., >50% or > 80%, as in KarMMA or CARTITUDE). The immunohistochemistry-based thresholds separate patients above or below the statistical median, but such cut-offs, and the absence of continuous monitoring of BCMA in the BM until confirmed relapse, make such approaches insensitive for evaluating impact of BCMA across the spectrum of lower surface densities that may be most relevant. In our trial, one of the patients who did not respond to treatment demonstrated no increase in BCMA density in the GSI run in (ABC = 89) (See patient 11; Figure 1). The reason for unresponsiveness is unknown; this patient did not harbor the recently described biallelic point mutation at the locus located on 16p13.13(23). In contrast, another patient, who was previously treated with BCMA CAR T cells on a trial using a different construct, and who progressed within 40 days on that trial, had a low level BCMA density (ABC = 199) at enrollment to our study, demonstrated a clear increase in BCMA ABC after GSI run in (ABC = 6072), and then responded to our treatment regimen (best response = VGPR) with a DOR of 8 months. Collectively, our findings indicate that BCMA density on malignant plasma cells is predictive of response and thus clinically relevant.

Amongst BCMA naïve patients, responses were deep and often durable; however, this was not observed in the BCMA treated populations, for several reasons. In one case, a patient who received a BCMA ADC, at the time of enrollment on the study, had negligible BCMA expression, and did not show any increase in BCMA ABC with the GSI run in. A second patient, who progressed by day 180 and had a very short duration of CAR T cell persistence (3.6% of peak level by day 28; and undetectable by day 60) had previously received BCMA CAR T cells generated with the same vector, raising the possibility that anti-CAR antibodies contributed to progression(24). Amongst the other 3 BCMA CAR T cell exposed patients, one died with a disseminated fungal infection at day 35 and the other two patients had a PFS of 309 days and 311 days.

While a BCMA CAR T cell dose response relationship has been reported by others, we found no association between the CAR T cell dose level and PFS or OS, suggesting the possibility that a GSI associated increase in target density abrogates the need for higher CAR T cell doses to achieve maximal efficacy. Further, despite anecdotal reports of a link between membrane bound BCMA level and CRS severity, we saw no correlation between BCMA ABC (following run-in) and CRS severity at all CAR T cell doses studied(25, 26). We did observe an association between development of ICANS and membrane bound BCMA levels following GSI run in, however 5/7 ICANS events were ≤ grade 2 and the median duration was 7 days. Nonetheless, the finding merits further investigation.

The development of uniform criteria to evaluate neurological events occurred after this trial opened. ASTCT consensus criteria for assessment of immune-effector cell mediated neurologic toxicity were incorporated into routine assessment following their publication and grading was retroactively applied. Overall, the incidence, duration, and severity of ICANS on our trial are difficult to compare across studies, complicated by differences in CAR product and the patient population enrolled. For example, disease burden, a factor associated with increased risk for developing ICANS and its severity in both BCMA and CD19 targeted CAR T-cell recipients, differs considerably between our trial and CARTITUDE participants. Utilizing criteria for high/intermediate/low tumor burden defined by the CARTITUDE investigators, our trial enrolled 41% with high tumor burden, 12% with intermediate, and 47% with low burden disease(27). The CARTITUDE trial, in contrast, had 18% high, 24% intermediate, and 65% low tumor burden(27). As noted in our results, all but one patient recovered fully from ICANS, and aside from the patient posthumously identified to have a concurrent fungal brain infection, there were no other neurologic DLTs. Despite our findings demonstrating overall safety associated with the GSI and BCMA CAR T cell combination, identifying differences in the frequency and severity of neurological toxicity requires larger randomized trials.

When compared to BCMA treatment naïve individuals, patients exposed to a BCMA targeting agent before study enrollment demonstrated less robust increases in BCMA density following GSI run-in. Nonetheless, the surface density for previously BCMA exposed patients still increased by a median 12-times (IQR 10 to 14.17, range 8.88 to 30.51) after 3 oral doses of GSI. The previously BCMA exposed patients represent a more heavily pretreated population, with a median 15 prior lines of therapy in comparison to 7 lines for BCMA naïve patients. Thus, while the difference in response to 3 doses of GSI is notable, determining if the cause was exposure to prior BCMA targeted therapy, or simply a function of the total number prior regimens, is not feasible from the current data.

To our knowledge, this is the first time that a potential role for the combination of GSIs with BCMA targeted CAR T therapy has been reported in humans. While no RP2D was established, we recommend proceeding with the 450 × 10⁶ dose level for future studies combining FCARH143 with a GSI and future clinical trials should explore alternative dosing strategies to investigate the therapeutic benefit that may accrue when BCMA target density is maximized through gamma secretase inhibition.

Research in Context.

Evidence before this study

We performed an exhaustive search employing ScienceDirect, Cochrane Library, Google Scholar and PubMed to identify any published data involving a BCMA targeting therapy administered in combination with a gamma secretase inhibitor (GSI). Key terms included: BCMA, CAR T-cell, immunotherapy, surface density, shedding, GSI, among others. The search covered reports from database inception through December 12, 2022. There are no prior published clinical studies combining a GSI with BCMA targeting therapy.

Added value of this study

To our knowledge, this trial is the first report to report the combination of any BCMA targeting therapy with a GSI to enhance anti-tumor efficacy. The approach represents an entirely novel approach to potentially overcome tumor escape through antigen loss. We report on safety of the combination and on the independent impact of GSI on BCMA density. The report is timely and relevant to the burgeoning field of BCMA targeting agents including CAR T-cells, bispecific T-cell engagers, antibody-drug conjugates and allogeneic CAR T-cell agents.

Implications of all the available evidence

The findings have significant implications for the field of antigen targeted immunotherapy for MM. The safety and feasibility of using a GSI to increase antigen target density supports extending the approach to other BCMA therapies and the results suggest exploration of alternate GSI dosing strategies are warranted to further improve the efficacy of BCMA targeting. In addition, the results support consideration of other approaches to enhance target antigen density more broadly to improve tumor specific targeting by immunotherapeutic agents.