Preclinical Assessment of CD171-Directed CAR T-cell Adoptive Therapy for Childhood Neuroblastoma: CE7 Epitope Target Safety and Product Manufacturing Feasibility

Neuroblastoma is the most common extracranial solid tumor of childhood with a heterogeneous clinical course (1). While neuroblastomas with favorable biology spontaneously regress or differentiate without therapeutic intervention, neuroblastomas with unfavorable biology often fatally progress despite intensive multimodal therapy (1–3). Maximally tolerated frontline intensive chemotherapy, radiation, consolidative autologous hematopoietic stem cell transplantation followed by retinoids and anti-GD2 antibody may cure up to 50% of high-risk patients. Accordingly, the development of new therapeutic modalities, which are tolerable in this patient population, is needed.

Immunotherapy is an attractive approach because it invokes immunologic effector mechanisms to which chemotherapy/radiation-resistant tumor cells are susceptible (4). In neuroblastoma, anti-disialoganglioside (GD2) antibodies have been most extensively utilized for antigen-specific immunotherapy; however, GD2 is not tumor specific and their antitumor effects are limited by passive biodistribution and the short half-life of the antibody (5–7). T cells expressing a chimeric antigen receptor (CAR) engage tumor cells independent of expression of HLA molecules and are activated via coordinated costimulation and CD3zeta signaling. A phase I study comparing EBV-cytotoxic T cells and peripheral blood T cells, both expressing first-generation GD2-specific CARs showed safety and transient responses (8). Further studies investigating the use of first-generation GD2 CAR expressing donor derived virus-specific cytotoxic T cells after allogeneic stem cell transplantation (clinicaltrials.gov ID: NCT01460901) or the use of third-generation (3G) GD2 CAR expressing T cells with inducible caspase-9 safety switch (clinicaltrials.gov ID: NCT01822652) are ongoing right now. Based on prior findings that the monoclonal antibody designated CE7 binds to an epitope on human L1CAM (CD171) in the context of tumor expression, a CE7 scFv was derived and assembled into a CAR for T cell–redirected tumor targeting (9, 10). While the molecular basis responsible for the tumor-selective CE7 epitope on L1CAM has not been fully elucidated, several lines of evidence suggest that binding is dependent on glycosylation (11–13). CD171 plays a role in oncogenesis as its expression correlates with tumor progression and metastasis in several solid tumors (14–18) and participates in the regulation of tumor cell differentiation, proliferation, migration, and invasion (14, 19–22). Initial assessment of target safety was made in a previously reported pilot clinical trial using autologous cloned CD8⁺ cytolytic T lymphocytes expressing a first-generation CE7-CAR (23). Six neuroblastoma patients were treated with doses up to 10⁹ cells/m² without obvious off-tumor toxicity. Not surprisingly, the duration and magnitude of persistence were limited presumably due to lack of CD4 help and costimulation. In an effort to increase therapeutic potency, we have engineered CE7-CARs containing a short spacer extracellular domain (24) and one [4-1BB; second-generation (2G)] or two (CD28 and 4-1BB; 3G) intracellular costimulatory signaling domains. A phase I clinical trial to determine the safety, feasibility, and optimal dose of 2G and 3G CE7-CAR T cells in patients with refractory or relapsed neuroblastoma has been initiated (Clinical Trial.Gov; IND FDA#16139) and we report the preclinical data of 2G CAR T cells manufactured from fresh autologous apheresis products below.

Research followed the tenets of the Declaration of Helsinki and was approved by the institutional review board of Seattle Children's Hospital (IRB#13740). Cancer Registry data (Oncolog) was searched for patients diagnosed with neuroblastoma between 2000 and 2014 in the Department of Pediatric Hematology-Oncology of Seattle Children's Hospital and surgical pathology material from 57 patients was reviewed. Paired tumor samples acquired at diagnoses and relapse (n = 7) were evaluated from 6 of the patients with recurrent disease and 1 patient with refractory disease. Data were retrospectively evaluated from medical records.

Two to three 1.5-mm cores from representative donor blocks were randomly distributed between recipient molds to create two TMAs.

A normal human and normal rhesus tissue array each containing 33 types of normal organs with 3 donors per tissue were purchased from US Biomax Inc.

Paraffin sections from arrays were placed on silanized slides and deparaffinized in xylene followed by 100% ethanol to water. Slides were quenched in 3% hydrogen peroxide, and then steamed in DIVA/buffer (pH 7.0; Biocare Medical) for 20 minutes to enhance antigen retrieval. Slides were subsequently incubated in Protein Block (Dako) for 5 minutes, then with CE7 mAb overnight at room temperature in a humid container. CE7 mAb was purified from CE7 hybridoma cells (kind gift of Dr. K. Blaser, University of Bern, Switzerland) utilizing the MAbTrap Kit (GE Healthcare, Bio-Sciences Corporation) per manufacturer's instructions. Slides were washed in Dako Buffer, incubated for 30 minutes in a Mouse Polymer Secondary reagent (Dako), washed again in Dako buffer and then incubated with the chromogen diaminobenzidine tetrahydrochloride, followed by hematoxylin counterstain. Normal tissues were scored as positive or negative. Neuroblastoma tissues were assessed by the percentage of tumor cells (<50%, 50%–90%, or >90%) showing positive membrane staining.

The CD17- specific CE7-CAR used herein was previously described (24). Briefly, the scFv was codon optimized and subsequently linked to a 12 AA spacer domain of the human IgG4 hinge followed by the transmembrane domain of human CD28 and a signaling module comprising the cytoplasmic domain of 4-1BB and CD3zeta (2G CAR). The cDNA clone was linked to a downstream T2A ribosomal skip element and truncated epidermal growth factor receptor (EGFRt) and cloned into the 3G SIN epHIV7 lentiviral vector transfer plasmid (25).

The lentiviral vectors were produced at the Center for Biomedicine and Genetics at City of Hope in Duarte (CA) under current good manufacturing practices (BB-MF 13830; Lentiviral Vector Manufacturing and Testing, City of Hope).

Aliquots from cryopreserved patient CE7-CAR T cell products were thawed, stimulated with irradiated peripheral blood mononuclear cells (PBMC), irradiated EBV-transformed lymphoblastoid cell lines (TMLCL) and OKT3 (30 ng/mL) and cultured in RPMI media (Gibco) supplemented with 10% fetal bovine serum (FBS; HyClone) and IL15 (0.5 ng/mL) plus IL2 (50 U/mL) for CD8⁺ cells, and IL15 (0.5 ng/mL) plus IL7 (5 ng/mL) for CD4⁺ cells. In vitro experiments were performed on day 11 after thaw. For in vivo experiments, cells were expanded until day 13, cryopreserved in aliquots as cell banks and thawed prior to injection.

PBMCs were isolated from adult Macaca mulatta blood by standard protocol using Ficoll. Cells were stimulated with anti-CD3 and anti-CD28 mAbs (BD Biosciences) and cultured in RPMI-1640 medium supplemented with 10% human AB serum (Gemini) or AIM V medium (Life Technologies) and IL2 (50 U/mL) as described (26, 27). Transduction was performed on day 1 using lentiviral supernatant (CE7 2G CAR), and cells were enriched for EGFRt expression by immunomagnetic selection. At the end of the stimulation cycle, cells were cryopreserved in aliquots as cell bank and thawed subsequently for in vitro expansion and infusion. CE7-CAR and mock control T cells were expanded using a rapid expansion protocol (28).

The neuroblastoma cell lines SK-N-Be2 and SK-N-DZ were obtained from the American Type Culture Collection (ATCC). The IL2 secreting, firefly luciferase (ffLuc) expressing SK-N-DZ was generated as described before (24). All neuroblastoma cell lines were cultured in DMEM (Cellgro) supplemented with 10% FBS and 2 mmol/L l-glutamine. TMLCLs and TMLCLs that expressed membrane-tethered CD3 epsilon-specific scFvFc derived from OKT3 mAb (TMLCL-OKT3; ref. 29) were cultured in RPMI 1640 supplemented with 10% FBS and 2 mmol/L l-glutamine.

Immunophenotyping was conducted with fluorophore-conjugated mAbs: CD3, CD4, CD8, CD45RA, CD45RO, CD62L, CD95, CD127, Tim3, PD-1 (BD Biosciences and Biolegend), and Lag3 (R&D Systems). Dead cells were excluded from analysis using a fixable viability stain (BD Biosciences). Cell surface expression of CD171 was analyzed using a fluorophore-conjugated mAb (Clone 014; Sino Biological) or the biotinylated CE7 mAb with a fluorophore-conjugated streptavidin secondary reagent. EGFRt expression was analyzed using a fluorophore-conjugated Cetuximab (Bristol-Myers-Squibb and BD biosciences). CAR surface expression was analyzed using a biotinylated antibody against the IgG Fab fragment with a fluorophore-conjugated streptavidin secondary reagent (Jackson Immuno Research and BD Biosciences). Flow analyses were performed on an LSRFortessa (BD Biosciences) and data were analyzed using FlowJo software (TreeStar).

Target cells were labeled with ⁵¹Cr (Perkin Elmer), washed, and incubated in triplicate at 5×10³ cells per well with T cells at various effector to target (E:T) ratios. Supernatants were harvested after a 4-hour incubation for γ-counting using Top Count NTX (Perkin Elmer), and specific lysis was calculated as previously described (9).

NOD/SCID/γc ^−/− mice. An NSG mouse tumor model was conducted under the Seattle Children's Research Institute Institutional Animal Care and Use Committee (IACUC)–approved protocol #13853. Adult male NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ [NOD scid gamma (NSG)] mice were obtained from The Jackson Laboratory or bred in-house. Mice were injected intracranially (i.c.) on day 0 with 2 × 10⁵ IL2-secreting, ffLuc-expressing SK-N-DZ tumor cells 2-mm lateral, 0.5-mm anterior to the bregma and 2.5-mm deep to the dura. Mice received a subsequent intratumoral injection of 2 × 10⁶ mock-transduced or CE7-CAR–modified T cells at a defined CD4:CD8 ratio of 1:1 after 7 days.

For bioluminescent imaging of tumor growth, mice received intraperitoneal (i.p.) injections of D-luciferin (Perkin Elmer; 4.29 mg/mouse). Mice were anesthetized with isoflurane and imaged using an IVIS Spectrum Imaging System (Perkin Elmer) 15 minutes after D-luciferin injection. Photon flux was analyzed within regions of interest using Living Image Software Version 4.3 (Perkin Elmer).

The study was performed according to recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The Institutional Animal Care and Use Committee approved the experiment protocol (UW#4159-01 and FHCRC#1638). Polyclonal 2G CE7-CAR T cells (1 × 10⁷/kg and 1 × 10⁸/kg) were infused intravenously (i.v.) in adult Macaca mulatta. Complete blood counts (CBC) and serum chemistry were measured in accredited laboratories. The in vivo persistence of transferred T cells in the peripheral blood was measured by flow cytometry using EGFRt as a marker gene (26, 27). Pre- and post-infusion plasma samples were examined for IFNγ, IL2, IL6, and TNFα, using a macaque-specific multiplex assay (Life Technologies).

A phase I clinical trial designed to assess the safety and feasibility of CE7-CAR T-cell therapy in children with relapsed or refractory neuroblastoma was opened to patient accrual in November 2014 (Clinical Trial.gov; IND FDA#16139) following IRB approval. Patients and their families provided written informed consent. Criteria for enrollment included patients: ≤18 years of age with high-risk neuroblastoma and insufficient response to first-line therapy (refractory) or progression during or after completion of initial therapy (relapsed); >7 days from prior myelosuppressive chemotherapy, >3 half-lives from tumor-directed antibody therapy, >6 weeks from prior radiopharmaceutical or myeloablative therapy, not receiving supraphysiologic doses of corticosteroids, not received prior genetically modified cellular therapy, adequate organ function, an absolute lymphocyte count ≥100 cells/μL, no active central nervous system disease including active seizures or other medical condition that would inhibit participation in the study. Patients could not be pregnant or breast-feeding, and all patients of childbearing potential must agree to use highly effective contraception.

Enrolled patients underwent standard apheresis for collection of approximately 5 × 10⁹ PBMCs. After isolation of a CD4⁺ and a CD8⁺ T-cell population using CD4 and CD8 magnetic beads (Miltenyi), cells were stimulated with anti-CD3/CD28 beads (Life Technologies) and cultured in X-Vivo media (Lonza) supplemented with 10% defined, irradiated, heat-inactivated FBS (GE Hyclone), IL15 (0.5 ng/mL), and for CD8⁺ cells IL2 (50 U/mL), for CD4⁺ cells IL7 (5 ng/mL). Transduction with the CE7 2G CAR lentiviral vector was performed on day 1 by centrifugation at 800 × g for 30 minutes at 32°C with lentiviral supernatant [approximate multiplicity of infection (MOI) = 0.1] supplemented with 1 mg/mL protamine sulfate (APP Pharmaceuticals). Residual stimulation beads were removed from culture prior to cryopreservation using the CTS magnet system (Life Technologies). Cells were expanded and cryopreserved in CryoStor CS-5 (Sigma) until the day of planned infusion. Aliquots were thawed subsequently for release tests, as well as in vitro and in vivo experiments.

Statistical analyses were conducted using Prism Software (GraphPad). Data are presented as means ± SD or SEM. Statistical analyses of survival were conducted by log-rank testing. Results with a P value less than 0.05 were considered statistically significant.

TMAs were generated using 73 tumor samples from 57 neuroblastoma patients and analyzed for CD171 expression using CE7 mAb IHC (Fig. 1A). The tumor samples were acquired at different time points (65% at diagnosis, 25% at second surgery, 10% at refractory/relapsed disease). The mean age of the patients at diagnosis was 3.4 (range, 0.1–18.7) years. The majority of the tumor samples were obtained from patients diagnosed with stage 4 (70%) and high-risk neuroblastoma (81%; Supplementary Table S2). CD171 expression was assessed by the percentage of tumor cells (<50%, 50–90%, or >90%) showing positive membrane staining (Fig. 1B). The intensity of positive staining in the neuroblastoma samples was uniform within and among tumors. All neuroblastoma displayed >50% CE7⁺ cells within the tumor sample, independent of risk status. In most samples, CE7 staining was evident in >90% of the cells [high risk = 83.8% (HR), intermediate risk = 100% (IR), low risk = 100% (LR); Fig. 1C]. Paired sample specimens from diagnostic and relapse biopsies were obtained from 7 patients. Specimens from 4 out of 6 patients retained >90% expression, 2 had decreased cell staining at relapse, and 1 had increased cell staining since time of diagnosis (Fig. 1D). Staining of bone marrow as a metastatic site of the neuroblastoma revealed CE7 staining in >90% of the neuroblastoma cells (Fig. 1E).

Prior studies have reported on the glycosylation dependent CE7 mAb epitope present on CD171 expressed by a variety of tumor types but not on CD171 expressed by nontransformed cells, such as peripheral blood monocytes (12). CE7 stains all CD171 tissues by IHC due to fixative denuding of carbohydrate moieties. In order to evaluate potential “on target” toxicities of CE7-CAR T cells in tissues known to express CD171, we performed IHC with the CE7 mAb on human tissue arrays. Consistent with published studies documenting CD171 expression on brain, peripheral nerves, kidney distal tubules, and other tissues, CE7 mAb stained known CD171 human tissues. Moreover, the extracellular domain of CD171 in rhesus macaques (NCBI: XP_001087861.1) is 100% homologous to that of humans (NCBI: NG_009645.3), and rhesus tissue array CE7 mAb IHC showed the same distribution of CD171 expression to that of humans (Fig. 2A). In rhesus, 28 out of 33 (84.8%) analyzed tissues did not express CD171, in human 28 out of 34 (82.4%; Supplementary Figs. S1 and S2).

Using clinical grade lentivirus vector, autologous 2G CE7-CAR⁺/EGFRt⁺ T cell lines were generated from two rhesus macaques. Polyclonally activated peripheral blood-derived CD3⁺ T cells were transduced, expanded in IL2, and immunomagnetically purified to homogeneous levels of EGFRt expression (Fig. 2B; ref. 30). Cell surface expression of the CAR itself was assessed by staining with an antibody against the IgG Fab fragment (Supplementary Fig. S3A). Rhesus macaque CE7-CAR T cell lines exhibited CAR-redirected CTL lysis of the human CD171⁺ neuroblastoma cell line SK-N-Be2 but not of the CD171⁻ human cell line (Fig. 2C; ref. 24).

To evaluate the safety of adoptive transfer of autologous CE7-CAR CTLs in rhesus macaques and analyze their persistence in vivo, 2G CE7-CAR T cells were infused intravenously at a dose of 10⁷ EGFRt⁺ T cells/kg (M.m. 1) and 10⁸ cells/kg (M.m. 2; Fig. 2D). Of note, these cell doses exceed by 10x- (M.m.1) and 100x- (M.m.2) the initial dose level proposed in the ENCIT-01 clinical trial. CD3⁺EGFRt⁺ CE7-CAR T cells were detectable in the blood through day 14 (M.m.1) and day 7 (M.m.2) post infusion (Fig. 2E). Animals were monitored clinically for fever, respiratory distress, loss of appetite, diarrhea and weight loss, and pre- and post-infusion blood samples for CBC, serum chemistry, and cytokine levels were obtained. No immediate or delayed clinical abnormalities due to toxicity occurred. Body weight, temperature, serum chemistry, and CBC remained within normal limits (Fig. 2F; Supplementary Fig. S3B; refs. 31–33). On day +1, animal M.m.1 showed a minimal increase in serum IL6 (8.7 pg/mL) and both animals showed a transient increase of INFγ serum levels (M.m.1 = 32.4 pg/mL and M.m.2 = 127.3 pg/mL; Supplementary Fig. S3C). No increase in plasma levels of TNFα or IL2 was observed (data not shown). No detectable neurologic side effects were evident and no change in kidney function were observed, these being tissues that express CD171 based on IHC tissue arrays.

T-cell subset and phenotype of the first 5 consecutively enrolled patients in ENCIT-01 was assessed by flow cytometry (Table 1). Analyses were performed on viable CD3⁺ lymphocytes after gating on CD4⁺ or CD8⁺ cells (Fig. 3A). For 3 out of 5 patients (S03, S04, and S05), the CD4⁺ repertoire mainly consisted of cells displaying a T_EM phenotype (mean = 64.9%) while the fraction of T_CM was generally low in all apheresis products (mean = 22.5%; Fig. 3B). The CD4⁺ repertoire of S01 and S02 revealed low levels of T_EM cells (mean = 24.4%) and large numbers of T_N cells (mean of S01 and S02 = 38.7% versus mean of S03, S04, and S05 = 8.6%). The same pattern (high levels of T_EM and low levels of T_CM cells) was seen for the CD8⁺ component of apheresis products. Further, the minority of T cells in these apheresis samples expressed the inhibitory receptor PD-1 (mean for CD4⁺ cells = 23.4%, mean for CD8⁺ cells = 18.5%) and were negative for Tim3 and Lag3, as detected by FACS analysis (Fig. 3C; Supplementary Fig. S4).

A phase I clinical trial of defined 1:1 ratio of CD4⁺:CD8⁺ CE7-CAR T cells has been developed to assess the safety and feasibility of adoptive therapy for children with relapsed or refractory neuroblastoma (Fig. 4A). We have performed apheresis, selection for CD4⁺ and CD8⁺ cells and lentiviral transduction for the first 5 consecutively enrolled patients. Four of the 5 patients CAR T-cell products were successfully manufactured and met release criteria (Supplementary Table S1). The CD4⁺ CAR T-cell products expanded on average 9-fold (range, 2.5–16.3) and, the CD8⁺ CAR T-cell products expanded on average 11-fold (range, 2.6–18.4) during in vitro culture over 10 to 15 days (Fig. 4B). The manufacturing failure observed in S04 was due to defective T-cell proliferation. Of note, this patient was apheresed following ¹³¹I MIGB therapy and was lymphopenic (ALC = 339 cells/mm³) at the time of the apheresis procedure.

Manufactured and cryopreserved CD4⁺ and CD8⁺ CE7-CAR products express EGFRt in approximately 80% of cells (Fig. 5A). Both CD4 and CD8 subsets underwent extensive differentiation to effector cells in vitro as evidenced by loss of CD45RA⁺ expression and enrichment for CD45RO⁺CD95⁺CD127⁻ effectors (Supplementary Fig. S4C and S4D). This differentiation was not so extensive to lose all CD62L expression, a marker associated with naïve and central memory T cells (Fig. 5B). CD4 products acquired cells that express PD-1 and Tim-3 without LAG-3 expression, whereas CD8⁺ products were enriched for cells expressing TIM-3 and LAG-3 with low frequencies of PD-1⁺ cells (Fig. 5C).

Next, we sought to determine if patient-derived CE7-CAR T-cell products display antitumor cytolytic activity. Using a 4-hour CRA, we observed lysis of the CD171⁺ neuroblastoma target cell line SK-N-DZ but not of the CD171⁻ cell line TMLCL for all 4 patient CD8⁺ T cell products, whereas CD4⁺ products exhibited minimal cytolytic activity (Fig. 5D). Maximal lytic outputs from these products were assessed using the universal target cell TMLCL-OKT3, a transfected LCL expressing the anti-CD3 scFv on its surface. To assess antitumor activity in vivo, we performed adoptive transfer experiments in NSG mice with established human neuroblastoma xenografts stereotactically implanted in the cerebral hemisphere. All SK-N-DZ tumor–engrafted mice treated with intratumoral injection of 2 × 10⁶ 2G CE7-CAR patient products at a defined CD4:CD8 ratio of 1:1 demonstrated tumor regression (Fig. 5E).

Despite many therapeutic advances, high-risk neuroblastoma remains the third most common cause of pediatric cancer death and alternative treatment modalities are urgently needed (34). CAR T-cell immunotherapy has been investigated in early-phase clinical trials with dramatic antitumor potency observed in patients treated with CD19-specific CAR T cells for B-cell malignancies, such as acute lymphoblastic leukemia and non-Hodgkin lymphomas (35–38). However, definitive clinical potency of CAR T-cell adoptive therapy in solid tumors remains to be seen. The challenges in solid tumor adoptive therapy are multifold and include the identification of safe targets amenable to CAR recognition, tumor heterogeneity, and the hostile immunologic microenvironment of solid tumors.

No antigen equivalent to the CD19 molecule has been identified in solid tumors. Consequently, antigen targeting is focusing on tumor-restricted epitopes rather than tumor-specific proteins on the surface of tumor cells such as unique epitopes that are generated due to posttranslational modification, genetic mutations, or altered splicing (39, 40). Here, we provide evidence that the epitope on CD171 recognized by CE7 mAb is homogeneously and abundantly present on human neuroblastoma cells, that CARs, derived from the CE7 Vh and Vl scFvs, are bioactive against neuroblastoma, do not exhibit demonstrable toxicities in a robust non-human primate (NHP) model, and can be manufactured under GMP from patient-derived apheresis products.

Target antigen density affects CAR T-cell therapy efficacy as an analysis of the kinetics of CAR T-cell–mediated killing revealed that the magnitude of tumor destruction was dependent not only on the presence of target antigens but also on the level of antigen expression (41). We demonstrate that CD171, as detected by our target epitope CE7, is highly and uniformly expressed in neuroblastoma, independent of disease-risk group, stage, or time of biopsy (primary vs. relapse), supporting the use of CD171-targeted immunotherapy.

CAR T-cell tumor specificity is desired to avoid toxicity associated with on-target but off-tumor antigen recognition. In the case of CD19-specific CAR T cells, the on-target off-tumor side effect of B-cell aplasia and the consequent effect on humoral immunity is considered to be tolerable and manageable (42). However, respiratory failure triggered by the recognition of low levels of antigen on lung epithelial cells by the Her2/neu-specific CAR T cells used in a trial for patients with colon cancer (43) is certainly an unacceptable side effect. Strategies described by other groups for targeting antigens expressed in healthy tissues are the use of lower doses of CAR T cells compared with other clinical trials (44) or the use of less-activated and short time persistent T cells due to expression of a first-generation CAR (8). Both studies showed no severe on-target off-tumor toxicities. Another strategy is choosing truly tumor-specific antigens, which are rare or, alternatively, tumor-restricted epitopes. The CAR used in our study contains an scFv derived from the CD171-specific murine CE7 hybridoma, specific for an epitope in the extracellular domain of CD171 present on cancer cells but not on normal CD171⁺ cells (45). Meli and colleagues found that inhibition of glycosylation in neuroblastoma cells leads to a loss of cell surface binding sites for the chCE7 mAb (11), suggesting that CE7 mAb recognizes a tumor-specific glycosylated epitope on CD171 not present on normal tissues. Hong and colleagues demonstrated that monocytes express CD171 using flow cytometry and a CD171-specific antibody UJ127. These monocytes failed to exhibit binding of the CE7 mAb (12). While these data support the contention that the glycosylation-dependent epitope on CD171 recognized by CE7 mAb is tumor restricted, the presence of this epitope on tissues that are known to express L1CAM such as CNS, peripheral nerves, and distal collecting tubules of the kidney cannot be excluded, particularly because tissue fixation for IHC strips carbohydrates and unmasks the CE7 epitope in IHC (author's unpublished observation).

Given the expression of CD171 on critical tissues and organs in humans and the technical obstacles to delineating the presence or absence of the CE7 epitope in vivo, we elected to evaluate the safety of 2G CE7-CAR T-cell adoptive transfer in an NHP model system previously described by Berger and colleagues (33). In the two animal doses with CE7-CAR T-cell products at doses 10 and 100× higher than the planned starting dose of our phase I trial, we saw no overt clinical toxicities in treated animals over the 7 to 14 days of engraftment. Plasma levels of IFNγ and IL6 were increased on day 1 after CE7-CAR T-cell infusion and cytokine levels normalized over time indicative of transient nonspecific activation of CAR T cells upon initial engraftment (33). Plasma cytokine levels in CE7-CAR–treated animals were significantly lower than those reported in ROR1-CAR–treated NHPs, wherein ROR1 antigen is present on a subset of B cells (33, 42). Although engraftment was only transient, likely due to transgene-specific immune responses, these data suggest that the NHP model is relevant for evaluating the safety of targeting the CE7 epitope on tumor CD171 and this model revealed no overt acute on-target off-tumor toxicity.

Because pediatric patients with high-risk neuroblastoma undergo dose-intensive upfront therapy cytotoxic chemotherapy, including myeloablative chemotherapy consolidation therapy requiring peripheral blood stem cell (PBSC) rescue, we sought to assess the peripheral blood T-cell repertoire of patients after relapse and salvage therapies, a point in time when apheresis will be performed to manufacture CAR T-cell products on the phase I study. Analysis of T cells present in apheresis products of 5 consecutively enrolled patients with ALCs ranging from 339 to 2068 cells/mm³ revealed heterogeneous subset frequencies among patients. For instance, 2 of the 5 patients (S03 and S04) had low frequencies of CD45RA⁺/CD62L⁺-naïve T cells and a preponderance of effector memory CD45RO⁺/CD62L⁻ precursors in both CD4 and CD8 repertoires. For all patients, frequencies of T cells expressing PD-1, TIM-3, and LAG-3 were modest. Assessment of S04's T-cell repertoire was notable for low frequencies of CD45RA⁺CD62L⁺CD127⁺ naïve T-cell precursors and low numbers of CD45RO⁺CD62L⁺ central memory precursors and conversely high frequencies of CD45RO⁺CD62L⁻ effector memory T cells. T_EM may be poor substrate cells for in vitro proliferation and have been documented to have limited persistence following adoptive transfer (26, 29, 46).

Finally, we demonstrated that T cells from 4 out of 5 heavily pretreated neuroblastoma patients were amendable to lentiviral vector transduction and competent for subsequent ex vivo proliferation to clinical cell doses. The final CD8⁺ CE7-CAR T-cell products expressed high levels of CD45RO and low levels of PD1. Furthermore, we demonstrated that these patient-derived products exhibit in vitro antitumor potency and in vivo antitumor effector functioning. However, one limitation of our study is that we do not describe the mechanism of resistance leading to disease progression in 2 out of 4 mice after initial tumor response. If antigen escape variants as seen in patients treated with CD19 CAR T cells (47) or antigen loss as seen in ovarian cancer bearing mice treated with CD171 CAR T cells (48) occur needs to be addressed in future experiments. The one patient who had insufficient T-cell ex vivo expansion underwent apheresis 8 weeks after I¹³¹-MIBG therapy, indicating a possible negative influence of ¹³¹I-MIBG therapy on T-cell growth that requires further analysis in a larger patient cohort.

In summary, our preclinical data demonstrate that CD171 is an attractive neuroblastoma antigen that can be targeted using the CE7-derived CD171-CAR T cells. A phase I study evaluating safety for this adoptive T-cell therapy in children with relapsed or refractory neuroblastoma is currently ongoing at Seattle Children's Hospital.

M.C. Jensen reports receiving commercial research grants from, holds ownership interest (including patents) in, and is a consultant/advisory board member for Juno Therapeutics; and reports receiving speakers bureau honoraria from Celgene. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Künkele, L.S. Finn, J.R. Park, M.C. Jensen

Development of methodology: A. Künkele, L.S. Finn, A.J. Johnson, O. Finney, M. Berger, J.R. Park, M.C. Jensen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Künkele, A. Taraseviciute, L.S. Finn, C. Berger, O. Finney, C.A. Chang, L.S. Rolczynski, C. Brown, S. Mgebroff, M. Berger, M.C. Jensen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Künkele, L.S. Finn, A.J. Johnson, C. Berger, O. Finney, C.A. Chang, M. Berger, J.R. Park, M.C. Jensen

Writing, review, and/or revision of the manuscript: A. Künkele, A. Taraseviciute, L.S. Finn, A.J. Johnson, C. Berger, C.A. Chang, C. Brown, S. Mgebroff, M. Berger, J.R. Park, M.C. Jensen

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Künkele, L.S. Finn, C. Berger, O. Finney, C.A. Chang, L.S. Rolczynski, S. Mgebroff, M. Berger, M.C. Jensen

Study supervision: A. Künkele, M.C. Jensen

Other (contributed to images): Laura S. Finn

The authors thank V. Hoglund for her support with the TMA staining.

A. Künkele was supported by the German Research Foundation (DFG, Deutsche Forschungsgemeinschaft). M.C. Jensen was supported by the SU2C/St. Baldrick's Pediatric Immunogenomics Dream Team Grant. The Jensen Laboratory at Ben Towne Center for Childhood Cancer Research is the recipient of financial support from the Ben Towne Pediatric Cancer Research Foundation, Make Some Noise: Cure Kids Cancer Foundation Incorporation, Seattle, Children's Guild Association, Andrew McDonough B+ Foundation, Journey 4 A Cure, and Zulily.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.