Autologous CD19-Targeted CAR T Cells in Patients with Residual CLL following Initial Purine Analog-Based Therapy

Abstract

Patients with residual chronic lymphocytic leukemia (CLL) following initial purine analog-based chemoimmunotherapy exhibit a shorter duration of response and may benefit from novel therapeutic strategies. We and others have previously described the safety and efficacy of autologous T cells modified to express anti-CD19 chimeric antigen receptors (CARs) in patients with relapsed or refractory B cell acute lymphoblastic leukemia and CLL. Here we report the use of CD19-targeted CAR T cells incorporating the intracellular signaling domain of CD28 (19-28z) as a consolidative therapy in 8 patients with residual CLL following first-line chemoimmunotherapy with pentostatin, cyclophosphamide, and rituximab. Outpatients received low-dose conditioning therapy with cyclophosphamide (600 mg/m²), followed by escalating doses of 3 × 10⁶, 1 × 10⁷, or 3 × 10⁷ 19-28z CAR T cells/kg. An objective response was observed in 3 of 8 patients (38%), with a clinically complete response lasting more than 28 months observed in two patients. Self-limited fevers were observed post-CAR T cell infusion in 4 patients, contemporaneous with elevations in interleukin-6 (IL-6), IL-10, IL-2, and TGF-α. None developed severe cytokine release syndrome or neurotoxicity. CAR T cells were detectable post-infusion in 4 patients, with a longest observed persistence of 48 days by qPCR. Further strategies to enhance CAR T cell efficacy in CLL are under investigation.

Graphical Abstract

Geyer et al. report the results of a phase I trial investigating CD19-targeted CAR T cells as consolidative therapy in patients with residual CLL following initial chemoimmunotherapy. Although this therapy was well tolerated, and 2 of 8 patients achieved a complete response, CAR T cell expansion was limited, likely, in part, because of suboptimal lymphodepletion.

Introduction

Chronic lymphocytic leukemia (CLL) remains the most prevalent leukemia among adults in the United States.¹ The natural history of CLL is considerably heterogeneous, and although many patients may be managed with an initial period of observation, the majority of patients will require therapy. In younger, medically fit patients without loss of chromosome 17p/TP53, standard initial therapy commonly consists of combination chemoimmunotherapy, often utilizing a purine analog-based regimen such as fludarabine (Flu), cyclophosphamide (Cy), and rituximab (FCR) or pentostatin, Cy, and rituximab (PCR).2, 3 Although these regimens are associated with overall response rates of approximately 90%, patients with unfavorable molecular features, including an unmutated immunoglobulin heavy chain variable (IgHV) gene, or those with persistent disease following initial purine analog-based therapy continue to exhibit a suboptimal duration of response when treated with standard chemoimmunotherapy.3, 4, 5 Therefore, patients with residual disease following initial purine analog-based therapy may benefit from novel approaches aimed at deepening the response.

We and others previously reported complete response (CR) rates of 20%–40% in patients with relapsed or refractory (R/R) CLL treated with second-generation CD19-targeted chimeric antigen receptor (CAR) T cells.6, 7, 8, 9, 10, 11, 12, 13 In contrast, CR rates exceeding 70% among patients with R/R B cell acute lymphoblastic leukemia (ALL) have been reported using second-generation CD19 CAR T cells.6, 14, 15, 16, 17, 18, 19, 20, 21, 22 In patients with R/R ALL, we observed that patients with only minimal residual disease (MRD) were more likely to achieve durable remissions and less likely to experience severe cytokine release syndrome (CRS) compared with those with morphologic disease at the time of CAR T cell therapy, suggesting that treatment in a state of low disease burden may be more efficacious and safer.²¹

We hypothesized that an overwhelming disease burden at the time of CAR T cell infusion could dampen the proliferative and cytolytic activity of CAR T cells in CLL. To this end, we conducted a phase I trial in previously untreated patients with CLL with residual disease following initial chemoimmunotherapy with six cycles of PCR to assess the efficacy of 19-28z CAR T cell therapy after initial disease debulking and low-dose conditioning chemotherapy (NCT01416974). Outpatients received escalating doses of 19-28z CAR T cells ranging from 3 × 10⁶ to 3 × 10⁷ CAR T cells/kg, administered outpatient. Here we report the safety and feasibility of this approach, demonstrate persistence of CAR T cells post-infusion, describe objective responses in a subset of patients treated with 19-28z CAR T cell therapy following initial PCR chemoimmunotherapy, and discuss strategies to overcome the limitations observed when utilizing this therapeutic approach.

Results

Patient Characteristics

The logistics of the clinical study procedures and patient eligibility are detailed in Materials and Methods and summarized in Figure 1. Eight eligible patients were enrolled after attaining a partial response (PR) to initial chemoimmunotherapy consisting of PCR. In all patients, PCR represented the first line of therapy administered for CLL. Clinical characteristics of treated patients are presented in Table 1. All 8 patients were men. The median age was 58 years at the time of CAR T cell infusion (range, 45–70). High-risk molecular and cytogenetic features included unmutated IgHV (n = 7) and deletion of chromosome 11q (n = 1). The median CLL international prognostic index (CLL-IPI) score was 5 (range, 2–6), with CLL-IPI scores of 2–3 and 4–6 corresponding to the intermediate and high risk categories, respectively.²³ Autologous T cell collection was performed at a median of 3.2 months (range, 2.3–7.0) following the last PCR chemoimmunotherapy, and CAR T cells were infused at a median 6.6 of months (range, 3.9–11.7) following the last chemoimmunotherapy. At the time of CAR T cell infusion, 5 of 8 patients had pathologically enlarged lymph nodes (>1.5 cm in the longest axis), one patient had already experienced overt progression of CLL following PCR on the basis of a briskly rising absolute lymphocyte count, and two patients exhibited an increase in lymph node size following PCR but did not meet the formal criteria for progression. The median bone marrow (BM) cellularity at CAR T cell infusion was 45% (range, 20%–60%); CLL comprised a median of 10% (range, 5%–50%) of BM cellularity upon morphologic evaluation of biopsy specimens.

Figure 1.

Study Design

(A and B) Overall flow of clinical study (A) and schematic of 19-28z gammaretroviral vector (B).

Table 1.

Demographic and Clinical Characteristics of Treated Patients and Outcomes

Pt	Dose Level (CAR T Cells/kg)	Age	Cyto	IgHV	CLL-IPI Score (When Starting PCR)	Time from PCR to CAR T Cell Infusion (Months)	Disease Status Post-PCR	BM Pre-CAR T Cell Infusion		Disease Status Pre-CAR T Cell Infusion	Infused CAR T Cell Dose per Kg	Best Response Post-CAR T Cell Infusion			Admitted with Fever?	Time to PD (Months)	Next Therapy
								Cellularity (%)	Extent of CLL (%)			BM	LN	Overall (IWCLL)
1	DL #1 3 × 106	59	NK	U	5	4	PR	20	10	stable PR	3.7 × 106	CCR	*	CCR	no	52.8	none
2		68	del13q	M	2	6	PR	20	20	stable PR with persistent LNs (2 cm max)	3.9 × 106	NR	*	SD	no	43.3+ (died without PD)	none
3		45	NK	U	5	6	PR	60	5	stable PR with persistent LNs (3 cm max)	3.6 × 106	nPR	NR	PD	no	3.2	ibrutinib
4	DL #2 1 × 107	56	NK	U	5	7	PR	60	50	PD w/ ↑ ALC	1.3 × 107	NR	NE	PD	yes	2.8	alloHSCT
5		66	NK	U	5	11	PR	40	10	stable PR with persistent LNs (2 cm max)	1.0 × 107	PR	*	SDa	no	14.5	BR
6		68	trisomy 12	U	6	12	PR	30	40	stable PR	0.9 × 107	CCR	*	CCR	yes	28.6	ibrutinib
7	DL #3 3 × 107	54	del13q	U	3	6	PR	50	5	mildly increasing LNs (4.8 cm max)	3.5 × 107	MRD+ CR	NR	PD	yes	2.9	BR ± idelalisib (study)
8		58	del11q	U	3	7	PR	50	5	mildly increasing LNs (7 cm max)	3.7 × 107	NR	NR	SD	yes	12.6	ibrutinib

Pt, patient; Cyto, cytogenetics; NK, normal karyotype; del, deletion; U/M, unmutated or mutated immunoglobulin heavy chain variable region; CLL-IPI, chronic lymphocytic leukemia international prognostic index; CR, complete response; CCR, clinical CR (defined as achievement of CR by all IWCLL criteria, including BM aspirate, but without trephine biopsy); PR, partial response; nPR, nodular PR; NR, no response; SD, stable disease; PD, progression of disease; DL, dose level; BM, bone marrow; LN, lymph nodes; NE, not evaluable; N/A, not applicable; MRD, minimal residual disease; BR, bendamustine and rituximab; alloHSCT, allogeneic hematopoietic stem cell transplantation; *Extent of adenopathy not evaluated following CAR T cell infusion because of a low burden of nodal disease prior to infusion.

^aAchieved PR in bone marrow without formal satisfaction of IWCLL critieria for PR or PD.

The median absolute lymphocyte counts (ALCs, K/μL) were 1.4 and 1.3 on the day of conditioning (Cy, 600 mg/m²) and on the first day of CAR T cell infusion, respectively. Compared with patients with R/R B cell acute lymphoblastic leukemia (B-ALL) undergoing conditioning with higher-dose Cy or Flu and Cy and those with R/R CLL receiving Flu and Cy conditioning prior to 19-28z CAR T cell administration in other clinical studies at Memorial Sloan Kettering Cancer Center (MSKCC), the patients in this study exhibited a minimal reduction in ALC and a higher ALC at the time of CAR T cell infusion (Figure 2).

Figure 2.

Lymphodepletion following Conditioning Chemotherapy

(A and B) Median absolute lymphocyte count (ALC) on the first day of CAR T cell infusion following conditioning chemotherapy (A) and median change in ALC from the beginning of conditioning chemotherapy to the first day of CAR T cell infusion (B) in patients enrolled on this trial compared with other clinical trials investigating 19-28z CAR T cells at MSKCC in patients with relapsed or refractory CLL (NCT00466531) or B-ALL (NCT01044069)21. Bars indicate interquartile range. Results are stratified by underlying disease status and conditioning regimen. R/R, relapsed or refractory; Flu, fludarabine; Cy, cyclophosphamide.

Leukapheresis and CAR T Cell Products

Median CD4:CD8 ratio in collected autologous T cells was 0.8:1 (range, 0.4:1–7.3:1). The median CAR T cell product manufacturing time was 11 days (range, 9–14). The median cumulative T cell expansion was 59.9-fold (range, 19.8–468.4) at the end of production (Figure S1). The median CD4:CD8 ratio in end of process (EOP) CAR T cells was 3.0:1 (range, 0.4:1–12.7:1). In all patients undergoing T cell collection following PCR chemoimmunotherapy, the target CAR T cell dose was successfully reached. Immunophenotypic characteristics of EOP CAR T cells are summarized in Table S1.

Safety and Toxicity

Conditioning chemotherapy and 19-28z CAR T cells were administered in the outpatient setting as described above. CAR T cell infusion was generally well-tolerated, with adverse effects summarized in Table 2. Following CAR T cell infusion, four patients, all on the higher two dose levels of the trial, were admitted on day +1 following the second infusion of CAR T cells (n = 2) or day +2 (n = 2) with temperatures of 38.0°C or higher. Peak temperatures in these four patients were 38.2°C–39.4°C, and the median number of days with fever was 1 (range, 1–2). The median total length of hospital stay was 3 days (range, 3–5). Admitted patients received supportive care; none had identifiable infection following evaluation. Fevers were attributed to grade 1 CRS in these 4 patients. No patients required tocilizumab, corticosteroids, or intensive care unit admission. No neurologic adverse effects were observed. No dose-limiting toxicities (DLTs) were observed at any of the 3 dose levels of CAR T cell infusion.

Table 2.

Adverse Effects Definitely, Probably, or Possibly Related to Protocol Therapy as Assessed by CTCAE v4.0

Toxicity	Total	Grade
		1	2	3
Fever	4	3	1	0
Thrombocytopenia	3	3	0	0
Lymphopenia	2	0	0	2
Chills	2	2	0	0
Hypocalcemia	1	0	1	0
Neutropenia	1	0	1	0
Leukopenia	1	1	0	0
Fatigue	1	1	0	0
Flushing	1	1	0	0
Hyperglycemia	1	1	0	0
Hypoglycemia	1	1	0	0
Hypernatremia	1	1	0	0
Hypoalbuminemia	1	1	0	0
INR increased	1	1	0	0
Nasal congestion	1	1	0	0
Pruritus	1	1	0	0
Vomiting	1	1	0	0
Weight loss	1	1	0	0

No grade 4-5 toxicities were observed.

CAR T Cell Persistence

CAR⁺ T cells were not definitively detected in the peripheral blood or BM of any patients by fluorescence-activated cell sorting (FACS; at a threshold of ≥1% of CD3⁺ cells) or by qPCR post-infusion. Rates of detection could be considerably increased after brief in vitro T cell expansion for 48% of the samples with sufficient leukocyte numbers. This modified assay, depicted in Figure S2, allowed detection of CAR T cells in 4 patients by qPCR (and by FACS in 2 of these 4). In 2 patients, leukocyte numbers were not sufficient for in vitro expansion. The maximal persistence of CAR T cells by FACS or qPCR was 34 and 48 days, respectively (Figure S3), using the modified assay. Greater CAR T cell persistence was not associated with an objective response (p = 1.0). B cell aplasia was not observed in treated patients.

Cytokine Levels

The levels of several immunoregulatory cytokines reliably peaked 2 days following the first day of CAR T cell infusion (designated day +2), including interleukin-2 (IL-2), IL-6, IL-10, IL-15, inducible protein-10 (IP-10), and transforming growth factor α (TGF-α) (Figure S4). A trend toward a greater increase in IL-6, IL-10, IL-2, and TGF-α levels from baseline levels to day +2 was observed among patients readmitted with grade 1 CRS (p = 0.06 for each) (Figure 3). Levels of cytokines associated with Th2-type immune responses (e.g., IL-4, IL-5, and IL-13) did not consistently rise in the days following CAR T cell infusion, and the extent of change was not significantly associated with the occurrence of fever (data not shown).

Figure 3.

Elevations in Cytokine Levels following CAR T Cell Infusion

Median fold change in cytokine levels (log10 scale) from prior to conditioning chemotherapy to day 2 post-infusion among patients admitted with fevers following CAR T cell infusion versus among patients not admitted with fevers. Differences in fold change in IL-6, IL-10, IL-2, and TGF-α between groups approached significance. Day 2 post-infusion cytokine levels were not available for one patient not readmitted with fever.

Clinical Responses

Clinical responses are summarized in Table 1. Objectives responses were observed in 3 of 8 patients (38%). Two patients (25%) achieved the best response of clinical CR (CCR), reflecting achievement of CR by International Workshop on Chronic Lymphocytic Leukemia (IWCLL) criteria, including BM aspirate but without a confirmatory BM biopsy core, and one (13%) exhibited BM PR based on regression of BM infiltrate compared with findings prior to CAR T cell infusion. Two additional patients developed overall progression of disease by IWCLL criteria but exhibited nodular PR (nPR) or minimal residual disease-positive (MRD+) CR within the BM.

With median follow-up time among survivors of 57.9 months (range, 40.2–69.0), median progression-free survival (PFS), measured from the first day of CAR T cell infusion as described above, was 13.6 months (95% confidence interval [CI; 2.8, 43.3]), and median overall survival (OS) was not reached (95% CI [34.9, not reached]) (Figure 4). PFS and OS at 3 years were 25% (95% CI [3.7, 55.8]) and 88% (95% CI [38.7, 98.1]), respectively. Median PFS from completion of PCR chemoimmunotherapy to progression of disease post-CAR T cell infusion was 22.7 months. Of the two patients achieving CCR, one experienced progression of CLL 28.6 months post-infusion and the other 52.8 months post-infusion. Among the 7 patients with disease progression, all except patient 1 received subsequent therapy (Table 1); one such patient died 34.9 months post-infusion after large cell transformation was noted. Another patient died suddenly and unexpectedly of suspected cardiac etiology 43.3 months post-infusion. An additional patient developing progression of disease post-CAR T cell infusion achieved subsequent disease control with ibrutinib but then developed progressive adenopathy, pathologically associated with increased Hodgkin-Reed-Sternberg (HRS)-like cells, Epstein-Barr virus (EBV)-negative, described as a “Hodgkin-like lesion” rather than the typical classical Hodgkin lymphoma variant of Richter transformation.²⁴ This patient received doxorubicin, bleomycin, vinblastine, dacarbazine (ABVD) for 6 cycles but exhibited persistent disease, and therapy was transitioned to rituximab and idelalisib, then to gemcitabine, vincristine, cyclophosphamide and prednisone on further progression, and most recently to bendamustine and obinutuzumab. Of the four living patients with progression of CLL post-CAR T cell infusion without transformation, one underwent allogeneic hematopoietic cell transplantation with achievement of durable CR, another received bendamustine and rituximab with a response observed, another received ibrutinib with an ongoing response observed, and one remains under observation.

Figure 4.

Survival Outcomes

(A and B) Progression-free survival (A, median 13.6 months) and overall survival (B, median not reached) following CAR T cell infusion.

An association between in vivo effector:target ratios and clinical response was not evident (Table S2). In all cases of disease progression, CD19 expression was preserved in the persisting CLL population, although it was described as dim in one patient relapsing following achievement of CCR.

Discussion

This phase I trial demonstrates the safety and feasibility of consolidative outpatient therapy with low-dose conditioning chemotherapy followed by 19-28z CAR T cells in patients with residual CLL after initial chemoimmunotherapy with PCR. There were no DLTs observed and minimal safety concerns at all 3 dose levels, suggesting a promising safety profile, although the highest planned dose level (3 × 10⁷ CAR T cells/kg) was not fully accrued. Four patients required brief hospital admission after developing fevers post-infusion, which were self-limited in all cases. No severe CRS and no neurologic adverse effects were observed, hematologic toxicity was transient, and other adverse events were largely mild. Three of 8 patients receiving 19-28z CAR T cells experienced an additional objective response beyond their initial response to PCR, including two patients with CCR. However, no objective response was observed in the remaining patients, and here we address several potential factors to account for the suboptimal activity observed and strategies for enhancing the antitumor efficacy of CAR T cell therapy in patients with CLL.

Several factors may account for the limited rates of objective response observed in the present report, most prominently lack of lymphodepletion following a single dose of Cy (600 mg/m²) as conditioning chemotherapy, with limited CAR T cell expansion observed in vivo. We hypothesized that patients with a lower burden of CLL following frontline PCR chemoimmunotherapy, compared with our prior study of patients with R/R CLL, would require a less intensive conditioning regimen of conditioning prior to CAR T cell infusion to achieve adequate CAR T cell expansion for a response. However, CAR T cells were detectable post-infusion in only 4 of 8 patients, even following brief ex vivo expansion, suggesting limited in vivo expansion and persistence in most treated patients. Other reports investigating CD19-targeted CAR T cells in B-ALL, CLL, and B cell non-Hodgkin’s lymphoma (B-NHL) have employed higher doses of Cy or incorporated Flu into conditioning chemotherapy.8, 9, 10, 11, 13, 15, 19, 25, 26 In the series reported by Kochenderfer et al.,8, 9 patients with R/R CLL received Cy (60–120 mg/kg) and Flu (25 mg/m²/day) for 5 days prior to CAR T cell infusion, a considerably higher total Cy dose than employed here, incorporating Flu for deeper lymphodepletion. Similarly, patients with R/R CLL in the report by Turtle et al.¹³ largely received Cy (30 mg/kg) and Flu (25 mg/m²/day) for 3 days as conditioning chemotherapy. Several series have now reported an association between incorporation of Flu into conditioning and more robust CAR T cell expansion and clinical response in patients with R/R B-ALL or B-NHL.19, 25, 26 The minimal change in ALC observed in this study following conditioning chemotherapy further suggests inadequate lymphodepletion (Figure 2). Patients with R/R CLL or with B-ALL receiving Flu and Cy conditioning prior to 19-28z CAR T cell infusion in other clinical studies at MSKCC have exhibited a greater reduction in ALC, a lower ALC at the time of CAR T cell infusion, and more consistent evidence of CAR T cell expansion in vivo.

We considered several alternative hypotheses regarding the suboptimal in vivo CAR T cell expansion observed. Exposure to PCR chemotherapy might have led to T cell dysfunction and poor expansion, although greater in vivo expansion of CAR T cells has been observed in other series enrolling patients with CLL previously treated with purine analog-based therapy.10, 13 Although the median total cumulative fold ex vivo expansion in this series was lower than those observed in another clinical trial at our institution (NCT0046653) investigating 19-28z CAR T cells in patients with R/R CLL (median 59.9-fold and 165.5-fold expansion observed, respectively), the total time in culture was shorter in this trial (median 11 versus 15 days), and the median cumulative fold expansion by 11 days in the other trial (38.1-fold) was similar to the total cumulative fold expansion observed here.¹² The infused CAR T cell dose might have been inadequate here; only two patients received the highest investigated dose of 3 × 10⁷ CAR T cells/kg because of slow accrual, particularly following the Food and Drug Administration (FDA) approval of ibrutinib. However, patients with R/R CLL have exhibited responses to CD19-targeted CAR T cells at doses lower than the maximal dose investigated here (3 × 10⁷ CAR T cells/kg). Specifically, CR has been observed in patients with R/R CLL following CAR T cell doses of 2 × 10⁵–2 × 10⁶/kg (in the series reported by Turtle et al.¹³), 2.5 × 10⁶–1.1 × 10⁷/kg (Kochenderfer et al.8, 9), and 0.14–11 × 10⁸ total CAR T cells (<3 × 10⁷/kg, Porter et al.¹⁰). The infused CAR T cell dose appears to be less strongly associated with a response in patients with CLL than in vivo expansion of the T cell product. CAR T cell products were skewed to an increased CD4:CD8 ratio (3.0:1), as we have previously reported in patients with R/R CLL for whom CAR T cells were manufactured using a similar process (CD4:CD8 ratio 10.5:1),⁷ although others have reported greater rates of objective response in patients with R/R CLL treated with CD4-predominant CAR T cell products.¹⁰ Finally, although most of the patients in the series here had unmutated IgHV (a poor prognostic feature), underlying disease biology alone seems insufficient to account for the limited antitumor activity observed in the present series. Responses have been observed in other series in patients with R/R CLL and high-risk cytogenetic and molecular features; 6 of 14 patients in the series reported by Porter et al.¹⁰ and 14 of 24 patients in the series reported by Turtle et al.¹³ exhibited deletion of chromosome 17p, with 9 of 14 patients in the former report having unmutated IgHV and 16 of 24 patients in the latter report demonstrating a complex karyotype as well.

In addition to poor in vivo CAR T cell expansion, the CLL burden at the time of CAR T cell infusion might have limited clinical responses as well. Among patients with B-ALL treated with 19-28z CAR T cells, we have identified a significant association between burden of disease at the time of treatment and durability of response.²¹ Others have additionally reported an association between nodal burden of CLL at the time of CD19 CAR T cell infusion and clinical response.7, 13 In the present report, autologous T cells were not collected until confirmation of residual disease, 2–7 months following completion of chemoimmunotherapy, and administration of consolidative CAR T cell therapy was therefore delayed, with actual infusions 3.9 to 11.7 months after completion of PCR. As described in Table 1, several patients were experiencing progressive adenopathy or rising ALC by the time of CAR T cell administration, which may have further limited efficacy; 19-28z CAR T cell therapy closer to the time of PCR completion, in a state of lower CLL burden might have enhanced activity. Although the effector-to-target ratios computed herein (Table S2) are comparable with ratios previously reported in patients with CLL who responded to CD19-targeted CAR T cell therapy,²⁷ several patients experienced further CLL progression following imaging studies. Additionally, the effector-to-target ratios estimated at the time of CAR T cell infusion do not account for subsequent in vivo expansion, which was limited in this series compared with others and would have influenced the effective in vivo CAR T cell dose. PCR has been chosen in this study given institutional familiarity and experience²⁸ as well as lower reported rates of infection and grade 3/4 hematologic toxicities, particularly neutropenia, and similar objective response rates among PCR versus FCR recipients in several series.3, 4, 29, 30 However, the more widely used FCR regimen (versus PCR) may also have led to deeper responses and more favorable in vivo effector-to-target ratios at the time of CAR T cell infusion.³¹ The median PFS from time of completion of PCR (22.7 months) in this series compared favorably with the duration of response among PCR recipients achieving PR in another large series (13.1 months),³ and three patients achieved progression-free survival more than 40 months from completion of PCR to eventual progression. However, the small size of the present series, lack of a randomized comparison with patients simply observed post-PCR, and non-progression of CLL in the absence of CAR T cell persistence limit our ability to suggest that consolidative CAR T cell therapy significantly altered the natural history of CLL in patients achieving PR to PCR.

The severity of CRS was universally mild in the present report, and no neurologic toxicity was evident. Four patients treated in the study here, all in the higher two dose cohorts, experienced grade 1 CRS, and none required tocilizumab or corticosteroids. Previous studies of second-generation CAR T cells have reported more frequent clinically severe CRS and neurologic toxicities, with severity of CRS associated with the infused dose of CAR T cells and in vivo expansion.7, 8, 9, 10, 11, 12, 13 Therefore, the limited CRS observed in this study is likely related, in part, to limited CAR T cell expansion. Additionally, T cells in patients with CLL exhibit significant proliferative and functional defects,³² which may account for tolerance of higher doses of CD19-targeted CAR T cells in patients with CLL compared with those with B-ALL without severe CRS. Nonetheless, we observed that the levels of IL-6, IL-10, and IL-2 reliably peaked on day +2 and a trend toward a significantly greater fold change in IL-6, IL-10, IL-2, and TGF-α from baseline in patients who experienced CRS, similar to other reported observations of an association between the rise of proinflammatory cytokines and CRS.10, 13, 15, 16, 18, 19, 25

This report and others investigating second-generation CD19-targeted CAR T cells in patients with CLL highlight the need for novel strategies to enhance CAR T cell expansion and activity in this setting. T cells from CLL patients exhibit a decreased proliferative capacity ex vivo in comparison with T cells from patients with B-ALL and multiple myeloma.³² CLL cells adopt various mechanisms to escape immune surveillance, including upregulation of inhibitory ligands (e.g., CD200, programmed cell death protein 1 ligand [PD-L1]), induction of CD8⁺ T cell exhaustion, suppression of natural killer (NK) cell cytotoxicity, and release of exosomes containing functional microRNA and proteins, promoting a cancer-associated fibroblast phenotype in stromal cells and supporting leukemic maintenance.33, 34, 35, 36, 37, 38 As described here, intensification of conditioning chemotherapy to optimize lymphodepletion, further reduction in CLL tumor bulk, particularly of nodal disease prior to CAR T cell infusion, or an increase in CAR T cell dose might enhance antitumor activity. However, other reports reflecting treatment of patients with R/R CLL using similar ranges of second-generation CAR T cell doses and more intensive conditioning chemotherapy have observed greater toxicity with similar or only slightly higher rates of CR.8, 9, 10, 11, 13 Therefore, a more efficient way to improve the efficacy of CAR T cells in CLL may involve further modification of CAR constructs to enhance anti-tumor potency or combination with immunomodulatory agents.³⁹ We have explored strategies to overcome CLL-mediated immune resistance in mouse models by further modifying CAR T cells to constitutively express an additional costimulatory ligand (e.g., 4-1BBL, CD40L) or to secrete immunoregulatory cytokines (e.g., IL-12, IL-18), and a phase I clinical trial of 19-28z/4-1BBL CAR T cells in patients with R/R CLL is ongoing at our institution.40, 41, 42, 43, 44, 45 Several other clinical investigations are ongoing to combine immunologic checkpoint blockade or the oral Bruton’s tyrosine kinase (BTK) inhibitor ibrutinib based on preclinical and preliminary clinical evidence showing that ibrutinib can enhance T cell proliferation, decrease PD-1 and CD200 expression, and increase skewing of the central memory phenotype of CAR T cells.12, 16, 32, 39, 46

In summary, administration of 19-28z CAR T cells to patients with residual CLL following initial chemoimmunotherapy with PCR was safe and well tolerated, and a subset of patients achieved a meaningful clinical response. The limited in vivo expansion and clinical efficacy of CAR T cells observed in the study are likely due to suboptimal timing of T cell collection and CAR T cell infusion in the setting of progressive CLL and inadequate lymphodepletion. Further studies are planned to optimize conditioning chemotherapy, disease setting, and CAR T cell design for use of this therapeutic modality in patients with CLL.

Materials and Methods

Clinical Study

We conducted a phase I clinical trial (NCT01416974) designed to assess the safety and maximum tolerated dose (MTD) of CD19-targeted CAR T cells as consolidative therapy following initial chemoimmunotherapy with 6 cycles of PCR in previously untreated patients with CLL.³ Patients with CLL with evidence of residual disease (PR, nPR, or CR with detectable MRD) upon BM examination and cross-sectional imaging studies following initial therapy with 6 cycles of PCR were eligible for enrollment and subsequent autologous T cell collection. Additional eligibility criteria included adequate organ function. The presence or absence of specific cytogenetic or molecular characteristics did not affect eligibility, although patients with Richter syndrome or a history of allogeneic hematopoietic cell transplantation (alloHCT) were considered ineligible. Patients achieving MRD-negative CR following PCR chemoimmunotherapy were ineligible for leukapheresis and CAR T cell infusion; one patient signed informed consent for this study but was ineligible for leukapheresis after achieving MRD-negative CR. MRD was assessed by PCR using patient-specific oligonucleotides amplifying the IgHV region locus and by multiparameter flow cytometry. BM examination was repeated prior to conditioning chemotherapy and CAR T cell infusion in treated patients. Treated patients received a single dose of Cy (600 mg/m² [day −2]) as conditioning chemotherapy, followed 2 days later by a split-dose infusion of escalating doses of CAR T cells (3 × 10⁶ [n = 3], 1 × 10⁷ [n = 3], and 3 × 10⁷/kg [n = 2]), with one-third of the total CAR T cell dose administered on day 0 and the remaining two-thirds administered on day 1 (Figure 1A). Patients received levetiracetam as seizure prophylaxis beginning 2 days prior to CAR T cell infusion. Cy and CAR T cell therapy was delivered in the outpatient setting.

The phase I study used a 3+3 design to evaluate the safety of three dose levels. The MTD was defined as the highest dose of CAR T cell infusion with a DLT rate of less than 33% of six patients. DLTs were defined as any of the adverse events listed in Table S3 occurring within 30 days from the last infusion of CAR T cells. The study was terminated early because of slow accrual after the second patient was enrolled at the highest dose level. Secondary objectives included assessment of disease response, evaluation of CAR T cell persistence, and characterization of changes in the cytokine tumor microenvironment. The institutional review board at MSKCC reviewed and approved this trial. All patients enrolled and treated in this trial gave written informed consent prior to participation. All clinical investigation was conducted according to the principles of the Declaration of Helsinki. Toxicities were assessed using the Common Terminology Criteria for Adverse Events (CTCAE) v4.0. CRS was graded according to the criteria in Table S4.

Generation and Expansion of Genetically Modified T Cells

Peripheral blood leukocytes were obtained from enrolled patients by leukapheresis, and CAR T cells were produced as described previously.7, 17, 47 Briefly, leukapheresis products were washed and cryopreserved. T cells from the thawed leukapheresis product were isolated and activated with Dynabeads ClinExVivo CD3/CD28 (Thermo Fisher Scientific) and transduced with gammaretroviral 19-28z vector stocks (Figure 1B). Transduced T cells were further expanded in the Wave bioreactor to achieve the target CAR T cell dose. EOP T cell products were characterized immunophenotypically by FACS using commercially available antibodies (Table S1).

Assessment of 19-28z CAR T Cell Persistence

Persistence of 19-28z CAR T cells in patient peripheral blood and BM was assessed by FACS with biotinylated goat anti-mouse immunoglobulin G (IgG) F(ab′)2 (Jackson ImmunoResearch) and by qPCR to determine vector copy number as described previously.7, 17

Some post-infusion samples were tested after brief expansion of T cells in vitro in the presence of Dynabeads ClinExVivo CD3/CD28.7, 17 In such patients, detection of CAR T cells was reported qualitatively because specific expansion of CAR T cells is not controlled.

Analysis of Cytokine Profiles following 19-28z Infusion

Serial serum samples were obtained before and after administration of Cy and following CAR T cell infusion. Cytokine profiles were analyzed using the Luminex FlexMAP 3D system and commercially available 38-plex cytokine detection assays as described previously.7, 17, 18

Response Assessment

Clinical responses were formally assessed using the IWCLL criteria at 3 months and 6 months following CAR T cell infusion on the basis of clinical examination, laboratory findings, BM aspirate, and biopsy and, where appropriate, cross-sectional imaging studies, including CT scans.⁴⁸ CCR was defined as achievement of CR by all criteria, including BM aspirate, but without trephine biopsy. Reduction in BM infiltrate or B-lymphoid nodules by 50% or more without full satisfaction of IWCLL criteria for PR or progression of disease was additionally considered to represent an objective response (BM PR), with official classification as stable disease (SD) by IWCLL criteria.

Statistical Considerations

PFS time was defined as the time from the first day of CAR T cell infusion to the date when progression was established by IWCLL criteria or the date of death in the absence of progression. PFS was computed separately to reflect the time from completion of PCR to the date of progression or death without progression. OS was calculated from the first infusion until death or the last follow-up. PFS and OS were estimated using Kaplan-Meier methods. The difference in maximal CAR T cell persistence between patients achieving versus not achieving an objective response was assessed using the Wilcoxon rank-sum test. Within the analysis of cytokine levels, the fold change (on a log₁₀ scale) from levels obtained pre-Cy to levels obtained on day +2 following CAR T cell infusion was compared between patients who were admitted with fevers versus not requiring admission following CAR T cell infusion using Wilcoxon rank-sum tests.

Author Contributions

M.B.G. performed the research, analyzed the results, and wrote the paper. I.R. performed the research and analyzed the results. B.S. performed the research, analyzed the results, and wrote the paper. X.W., Y.W., T.J.P., E.H., and N.L. performed the research. M.H. and S.M.D. analyzed the results. J.R. performed the research and analyzed the results. M.S. designed the research. J.H.P. and R.J. B. designed and performed the research, analyzed the results, and wrote the paper. All authors critically reviewed the paper.

Conflicts of Interest

M.B.G. receives honoraria from Dava Oncology. J.H.P. has a consulting/advisory role with Amgen), a consulting/advisory role with Juno Therapeutics and receives research funding from Juno Therapeutics, and receives research funding from Genentech/Roche. I.R. and R.J.B. have consulting/advisory roles, stock ownership, and royalty sharing agreements with and receive research funding from Juno Therapeutics. M.S. has a consulting/advisory role, royalty sharing agreement, and stock ownership with Juno Therapeutics.