Targeting refractory/recurrent neuroblastoma and osteosarcoma with anti-CD3×anti-GD2 bispecific antibody armed T cells

Abstract

Background

The survival benefit observed in children with neuroblastoma (NB) and minimal residual disease who received treatment with anti-GD2 monoclonal antibodies prompted our investigation into the safety and potential clinical benefits of anti-CD3×anti-GD2 bispecific antibody (GD2Bi) armed T cells (GD2BATs). Preclinical studies demonstrated the high cytotoxicity of GD2BATs against GD2+cell lines, leading to the initiation of a phase I/II study in recurrent/refractory patients.

Methods

The 3+3 dose escalation phase I study (NCT02173093) encompassed nine evaluable patients with NB (n=5), osteosarcoma (n=3), and desmoplastic small round cell tumors (n=1). Patients received twice-weekly infusions of GD2BATs at 40, 80, or 160×10⁶ GD2BATs/kg/infusion complemented by daily interleukin-2 (300,000 IU/m²) and twice-weekly granulocyte macrophage colony-stimulating factor (250 µg/m²). The phase II segment focused on patients with NB at the dose 3 level of 160×10⁶ GD2BATs/kg/infusion.

Results

Of the 12 patients enrolled, 9 completed therapy in phase I with no dose-limiting toxicities. Mild and manageable cytokine release syndrome occurred in all patients, presenting as grade 2–3 fevers/chills, headaches, and occasional hypotension up to 72 hours after GD2BAT infusions. GD2-antibody-associated pain was minimal. Median overall survival (OS) for phase I and the limited phase II was 18.0 and 31.2 months, respectively, with a combined OS of 21.1 months. A phase I NB patient had a complete bone marrow response with overall stable disease. In phase II, 10 of 12 patients were evaluable: 1 achieved partial response, and 3 showed clinical benefit with prolonged stable disease. Over 50% of evaluable patients exhibited augmented immune responses to GD2+targets post-GD2BATs, as indicated by interferon-gamma (IFN-γ) EliSpots, Th1 cytokines, and/or chemokines.

Conclusions

This study demonstrated the safety of GD2BATs up to 160×10⁶ cells/kg/infusion. Coupled with evidence of post-treatment endogenous immune responses, our findings support further investigation of GD2BATs in larger phase II clinical trials.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Targeting GD2 ganglioside on the surface of neuroblastoma cells with monoclonal antibodies prolonged progression-free survival but still was not curative in patients with high-risk neuroblastoma. Activated T cells armed with chemically conjugated bispecific antibodies were shown to be clinically safe and elicit secondary immune responses in adult patients with pancreatic, colon, breast, and prostate cancers.

WHAT THIS STUDY ADDS

• Using armed T cell immunotherapy platform against GD2-positive pediatric malignancies is safe and shows evidence of clinical activity and vaccine-like action in stimulating endogenous lymphocyte T cell and NK responses against GD2-positive cell lines in patients with neuroblastoma and osteosarcoma.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study adds another safe and promising armed T cell immunotherapy approach to be potentially used in GD2-positive malignancies such as neuroblastoma, osteosarcoma and melanoma.

Introduction

Despite aggressive treatments including intensive chemotherapy, autologous stem cell transplant (SCT),1 2 and anti-GD2 monoclonal antibodies (mAbs) therapy, a significant number of children with high-risk neuroblastoma (NB) and osteosarcoma (OST) develop recurrent disease3 4 which is associated with dire outcomes.1 5 6 There have been no new approved drugs for OST since the 1980s.7

GD2, highly expressed on nearly all NB and 80% of OST8 9 cases, is a target for antibody-dependent cellular cytotoxicity. Anti-GD2 mAbs are effective against minimal residual disease in NB.2 10 11 We hypothesized that targeting NB and OST with activated autologous T-cells (ATC) armed with GD2Bi (GD2BATs) could be a feasible, safe, and effective approach. Our preclinical studies demonstrated that GD2BATs potently kill GD2+tumor targets, proliferate, and induce release of Th₁ type cytokines on engagement of GD2+tumor targets.12 Redirecting non-major histocompatibility complex (MHC) restricted cytotoxicity of ATC with bispecific antibody (BiAb) to various cancers has provided encouraging clinical and anti-tumor immune responses in breast,13 14 prostate cancer15 16 pancreatic cancer,17 non-Hodgkin’s lymphoma,18 and multiple myeloma19 in our previous clinical trials. Infusions of BiAb armed ATC (BATs), combined with immune adjuvants such as granulocyte macrophage colony-stimulating factor (GM-CSF) and interleukin-2 (IL-2) to support ATC and boost immune responses,13–15 20 21 triggered endogenous responses. These responses included innate natural killer cell (NK) activities and cytokine/chemokine responses, suggesting in vivo immunization of patients against their own tumor-associated antigens.13 17 22

In this study, we evaluated a combination immunotherapy approach consisting of GD2BATs,12 23 interleukin-2 (IL-2), and GM-CSF in a phase I/II trial. Our primary objectives were to assess safety, estimate efficacy, and evaluate immune responses. The reported feasibility and safety profile, coupled with evidence of therapy-induced immune responses and clinical activity, provides a strong rationale for advancing the GD2BAT approach in NB.

Material and methods

Production of GD2BATs

Manufacturing of GD2Bi and GD2BATs occurred under BB-IND #15827, sponsored by L.G. Lum, at the KCI cGMP facility, Detroit, MI and the UV Center for Human Therapeutics in Charlottesville, VA under FDA Drug Masterfile File #19 248. Clinical-grade naxitamab (humanized 3F8) was chemically heteroconjugated to clinical-grade anti-CD3 mAb (OKT3; Miltenyi, Auburn, California, USA) to produce GD2Bi. Patient-derived lymphocytes (from peripheral blood mononuclear cells (PBMCs)) were collected through apheresis, activated with 20 ng/mL of OKT3, expanded using 100 IU/mL of IL-2, and cultured at a concentration of 1×10⁶ mononuclear cells/mL in RPMI medium supplemented with 2% human serum and 1% L-glutamine for 14–18 days. The ATC were then harvested, armed with 50 ng of GD2Bi per 10⁶ ATC to create GD2BATs, divided equally, and cryopreserved. Thawed product underwent quality control assessments, including viability, proportion of CD4+ and CD8+ cells, in vitro cytotoxicity against KCNR GD2+NB cell line (by incubating patients’ PBMCs with targets), testing mycoplasma, bacterial and fungal contamination testing, and endotoxin analysis before release for infusion.

Human subjects

Patients were enrolled in a phase I/II clinical trial at CHM, UV, and MSK between November 2013 and January 31, 2022. The protocol (ClinicalTrials.gov identifier NCT02173093) was reviewed and approved by the FDA, protocol review committees, and institutional review boards at respective institutions. Signed informed consent was obtained from all guardians/patients before enrolment. Eligible patients were aged ≤30 years at the time of enrolment and had recurrent or refractory NB, OST, or desmoplastic small round cell tumors (DSRCT). The phase II portion was restricted to patients with NB. Depending on the tumor type, patients were to have confirmation of evaluable or measurable disease by bone marrow (BM) studies, CT, anatomical imaging, and/or iodine-123 metaiodobenzylguanidine (MIBG). Adequate performance score (Lansky or Karnofsky score ≥70) and adequate organ function (total bilirubin ≤1.5 times the institutional upper limit of normal (ULN), ALT ≤3 times the ULN, creatinine ≤1.5 times the ULN, absolute neutrophil count ≥500/mm³, absolute lymphocyte count ≥500/mm³, and unsupported platelet count ≥50 000/mm³, and a cardiac shortening fraction ≥28% by echocardiogram) were prerequisites. A minimum of 2 weeks from the last chemotherapy and immunotherapy and 1 week from hematopoietic growth factors were required for enrolment.

Patients who had been previously treated with anti-GD2 antibodies were eligible. Date of data cut-off for this analysis was January 31, 2022.

Clinical trial design

The phase I study followed a standard 3+3 dose escalation design with dose levels of 40, 80, and 160×10⁶ GD2 BATs/kg/infusion twice per week, totaling 0.32, 0.64 or 1.28×10⁹ cells/kg in 8 divided doses. Low-dose IL-2 (300,000 IU/m²/day) and biweekly subcutaneous GM-CSF (250 µg/m²/dose) were administered starting 3 days before the first GD2BAT infusion and ending 1 week after the last infusion (figure 1A).

Figure 1.

(A) Protocol schema. It shows the treatment schema for the phase I/II study. (B) Survival curves for phase I and phase II. It shows the Kaplan-Meier overall survival curves for phase I (blue, OS=18.0 months), phase II (red, OS=31.2 months), and all patients (green, OS=21.0 months). GM-CSF, granulocyte macrophage colony-stimulating factor; OS, overall survival.

The phase I study aimed for a maximum of 18 patients, with an additional 22 patients required for the phase II cohort to confirm toxicity and estimate clinical efficacy. Primary endpoints included safety (phase I) and overall response rate (ORR, partial response (PR)+ complete response (CR)) for phase II. Secondary endpoints included overall survival (OS), immune responses, persistence of infused cell, and early fluorodeoxyglucose positron emission tomography (FDG PET) response postinfusions. Bridging chemotherapy with a 2-week washout was permitted in the interval (up to 90 days) between apheresis and initiation of protocol therapy, with a 6-week washout required If anti-GD2 mAbs were administered during bridging. Protocol eligibility was reaffirmed through laboratory testing, imaging and BM evaluation as needed before commencing protocol therapy.

Toxicity assessment

Toxicity, assessed using Common Terminology Criteria for Adverse Event (CTCAE) V.5.0, additionally incorporated grading of cytokine release syndrome (CRS) per the American Society for Transplantation and Cellular Therapy consensus criteria (ASTCT).24 Grade 3 chills, fever, headache, nausea, emesis, and hypotension lasting <72 hours were excluded from dose-limiting toxicities (DLTs) based on prior experience with BATs.13 17–19 22 25 Grade 3 fever, nausea, or vomiting was defined as DLTs only if persisting >72 hours. Patients were monitored twice weekly during BAT infusion days for toxicity evaluation.

Response evaluation

Disease evaluation in patients with NB involved Imaging and BM testing at enrolment, after bridging therapy (prior to protocol therapy), and 12 weeks after the first GD2BAT infusion, graded using the International Neuroblastoma Response Criteria (INRC).26 For non-NB tumors, radiological response was assessed using RECIST criteria.

Correlative studies

GD2BAT persistence studies and immune evaluations

To assess the BATs persistence was assessed by staining patient PBMCs with anti-mouse:IgG2a antibody (present on OKT3)13 or anti-idiotypic antibody A1G4 specific for hu3F8 at 2 hours post-GD2 BAT infusions and 1 week after the last infusion.

Specific cytotoxic lymphocyte (CTL) activity was evaluated using interferon-gamma (IFN-γ) enzyme-linked immunosorbent spots (ELISpots) induced by GD2-expressing NB (KCNR) and OST (MG-63) cell lines and natural killer (NK)-targeted cell line K562.12 Cytokines IL-6, IL-10, TNF-α, MIP-1β, and IP-10 were assessed by Luminex.22 Selected patients with measurable soft-tissue lesions underwent FDG PET/CT at baseline and between infusions #5 and #8.

Statistical analysis

Immune evaluations included evaluating means, SD, medians, and the distributions of the data to ascertain whether normal theory methods are appropriate. Paired t-test or Wilcoxon signed-ranks test were used for comparative analyses between each preimmunotherapy and postimmunotherapy (IT) time point for each patient. With a single-stage phase II trial design, it was estimated that a cohort of 22 phase II patients would have 80% power at a significance level of 0.05 to detect 25% difference in response rate compared with a historical response rate of 15% or less. Due to slow recruitment, the phase II portion of the trial enrolled half of the planned patients (12 of planned 22 enrolled and 10 completed therapy), and responses were analyzed descriptively without applying initially planned statistical methods in the study protocol.

Results

Patient characteristics

Between November 2013 and December 2017, 12 patients (7 NB, 3 OST and 2 DSRCT) were enrolled in the phase I study, with 9 completing therapy. 12 additional patients with NB were enrolled in the phase II portion, and 10 completed the protocol therapy. Table 1 provides a summary of patient characteristics, diagnoses, prior therapy, product characteristics, and survival of the phase I/II patients.

Table 1.

Diagnosis, prior therapy, product characteristics, and survival

Study phase/ dose level	ID	Diagnosis	Prior therapies	Total dose level/ dose×109/kg	Weight	Total dose (×109)	TX completed	Product
								Viability (%)	Cytotoxicity (%)	Cd3 (%)	Cd8 (%)	Cd4 (%)	Response end of therapy	OS (days)	Status
PI/1	IT20107	OST	CT: MAP, HD Ifosfamide	0.32	60.0	19.04	Yes	73	25	92	57	40	PD	75	DOD
PI/1	IT20108	OST	CT: MAP, IE	0.32	55.0	17.76	Yes	75	10	88	57	38	PD	139	DOD
PI/1	IT20111	NB	CT: SCT, Topo/ Cy, RT	0.32	62.9	20.08	Yes	79	53	98	31	61	SD	1476	DOD
PI/1	IT20113	NB	CT German NB 2004, SCT, multiple lines, sirolimus	0.64	19.4	NA	No	45	NA	NA	NA	NA	NA	92	DOD
PI/2	IT20115	OST	Multiple lines CT (MAP, IE, gemcitabine, docetaxel), RT	0.64	66.9	42.8	Yes	74	19	99	49	51	SD	593	DOD
PI/2	IT20116	DSRCT	CT, pazopanib	0.64	67.4	43.1	Yes	92	33	98	64	37	PD	201	DOD
PI/2	IT20119	NB	ANBL12P1, SCT, vaccine, m3F8, irino/TMZ	0.64	29.3	18.75	Yes	77	3	96	6	93	PD	590	DOD
PI/3	IT00010	NB	ANBL12P1, SCT, hu3F8, RT, vaccine	1.28	18.0	23.04	Yes	91	31	95	84	22	PD	494	DOD
PI/3	IT20122	NB	ANBL0532, SCT, RT, Ch14.18	1.28	25.0	32	Yes	86	21	83	31	42	PD	633	DOD
PI/3	IT00003	DSRCT	CT, Surg, RT	1.28	91.7	37.8	No	95.8	5	94	43	78	PD	83	DOD
PI/3	IT00005	NB	ANBL12P1, MSK13-260(m3F8), CT, MSK12-230 (Hu3F8), RT,	1.28	23.3	29.9	Yes	93.6	10	98	43	57	SD	1571	AWD
PII	IT00013	NB	ANBL0532, SCT, RT, Ch14.18, CT	1.28	51.4	88.96	Yes	88.6	47.9	97	52	49	PR	1640	NED*
PII	IT00017	NB	ANBL12P1, SCT, RT, ANBL 0032 (ch14.18), CT, Hu3F8	1.28	75.8	78.40	Yes	93.8	17.4	94	50	47	SD	1405	AWD
PII	IT00019	NB	CT, Hu3F8, RT, SCT+Haplo, NK cells	1.28	28.5	34.24	Yes	92.1	21.1	84	48	54	SD	935	DOD
PII	IT00022	NB	German NB 2004, MIBG, Hu3F8, Multiple CT	1.28	37.6	48.00	Yes	88.4	68.1	87	48	46	SD	614	DOD
PII	IT00023	NB	ANBL12P1, Ch14.18, MIBG, Hu3F8, vaccine, Hu3F8, Donor NK	1.28	25.2	25.60	Yes	83	45.7	97	45	44	PD	220	DOD
PII	IT00028	NB	CT, RT, Hu3F8, Vaccine, MIBG,	1.28	32.6	41.60	Yes	94.1	47.9	83	46	52	PD	709	DOD
PII	IT00031	NB	ABNL0532, tandem SCT, RT, Ch14.18, irino/TMZ/Ch14.18	1.28	56.6	59.04	Yes	96.9	24.8	90	63	39	PD	1460	AWD
PII	IT00033	NB	ANBL12P1, SCT, RT, ANBL0032, Ch14.18; topo/cy	1.28	39.2	50.16	Yes	86.5	35.1	96	86	19	SD	1218	AWD
PII	IT00040	NB	ANBL0532, tandem SCT, RT, Ch14.18	1.28	22.0	28.00	Yes	94.1	43.1	98	68	37	SD	902	DOD
PII	IT00084	NB	ANBL12P1, SCT, RT, ANBL1221 (irino/TMZ, Ch14.18)	1.28	67.2	69.04	Yes	84.2	12.1	98	40	59	PD	439	AWD

*After disease progression at 6 months and then receiving several additional lines of therapy.

ANBL1221, COG NB protocol for recurrent /refractory disease; AWD, alive with disease; ch14.18, chimeric GD2 mAb; COG, Children’s Oncology Group; CT, chemotherapy; DOD, died of disease; DSRCT, desmoplastic small round cell tumor; GM-CSF, granulocyte macrophage colony-stimulating factor; hu3F8, humanized 3F8 GD2 mAb; IE, ifosfamide and etoposide; irino/TMZ, irinotecan and temozolomide; MAP, methotrexate, doxorubicin, cisplatin; m3F8, murine 3F8 GD2 mAb; NB, Neuroblastoma; OS, overall survival; OST, Osteosarcoma; PD, Progressive disease; PI, phase I; PII, phase II; RT, radiation therapy; SCT, autologous stem cell transplant; topo/cy, topotecan and cyclophosphamide.

Reasons for not receiving therapy or not completing eight infusions in the phase I included failure to grow adequate ATC in one NB patient, rapid disease progression before GD2BATs in another NB patient, and rapid disease progression leading to death after the first three GD2BATs infusions in one patient with DSRCT. Additionally, two phase II patients did not undergo T cell collection and therapy because of rapid disease progression post-enrolment. All patients were heavily pretreated, with OST and DSRCT patients having 2–3 and 1–2 lines of prior therapy, respectively, and all phase I NB patients undergoing prior myeloablation followed by autologous stem cell transplant (SCT), with 14 NB patients having prior anti-GD2 antibodies. Table 2 details the distribution, disease status, prior therapy, and disease sites.

Table 2.

Distribution of patients, disease status, types of prior therapies, and disease involvement*

	Phase I	Phase II
Male-to-female ratio	7–2	5–5
Age, median (range) at enrolment, years	12.5 (5–30)	13.5 (6–25)
Diagnosis
Neuroblastoma  Osteosarcoma  DSRCT	5 3 1	10
Neuroblastoma disease status
Recurrent  Primary refractory	3 2	7 3
Prior therapies neuroblastoma
Autologous stem cell transplant  External beam radiotherapy  131I-MIBG  Anti-GD2 mAb  Other Phase I	5 5 2 4 3	6 7 3 10
Prior therapies osteosarcoma and DSRCT (n=5)
Median number of prior chemotherapy regimens	3
Disease involvement at study entry, neuroblastoma, n
Bone marrow  Skeletal MIBG avid lesions  Measurable disease (soft tissue, lymph nodes)	7 3 7 4	10 3 8 3
Disease involvement at study entry, osteosarcoma (n=3), DSRCT (n=2)
Bones  Lungs  Liver  Abdomen	2 2 1 2

*Three phase I and two phase II patients who were not treated (4) or did not complete therapy (1) were not reported in table 2.

DSRCT, desmoplastic small round cell tumors; MIBG, metaiodobenzylguanidine.

Preparation of cells and cell product characteristics

Cell products were successfully manufactured in 20 of 21 patients who underwent apheresis. For one patient with NB (IT20113), viability of the manufactured product (45%) was below the target (viability release ≥70%), and the minimum target cell number was not met. The remaining 20 patients met cell expansion goals and release criteria, with viability ranging from 73% to 96% (median 82.5%). CD3+, CD8+, and CD4+T cells percentages, as well as ⁵¹Cr release specific cytotoxicity against GD2+NB cell lines, are presented in table 1. CD8/CD4 ratios in the manufactured products ranged from 4:1 to 1:15.

Toxicity

In phase I study, 9 patients received a total of 72 GD2BAT infusions. No patients had treatment stopped due to adverse effects (table 3).

Table 3.

incidence and grade of toxicities by dose level

			Total # of episodes by grade
Dose level	Term	# patients experiencing	1	2	3	4
Level 1	Nausea	3	6	1
(n=3)	Diarrhea	2	2
	Chills	1	1	6
	Fatigue	2			2
	Headache	2	4	4	3
	Fever	1		2	3
	Hypotension	1		1
	Insomnia	1		1
	Anorexia	2	1	1	1
	Pain	2	1	1	1
Totals			15	17	10
Level 2
(n=3)	Nausea	3	3	4	2
	Anorexia	3	6	3	3
	Fever	3	1	6	4
	Emesis	1			1
	Chills	2		3	4
	Headache	3	2	2	9
	Pain	3		6
	Hypotension	1		1
	Fatigue	3	3	2	6
	Infection, tumor site	1		1
Totals			8	27	19
Level 3
(n=3)	Nausea	2	1	1
	Anorexia	0
	Fever	2	6	2
	Emesis	2		3
	Chills	0
	Headache	2	3	2
	Pain	1	2
	Hypotension	1	3	5
	Fatigue	0
	Anemia	1			1
	Sore throat	1	1
	Sinus Tachycardia	1	3
Totals			19	13	1
Phase II
(n=10)	Nausea	3	2	3
	Anorexia	0
	Fever	4	10	4
	Emesis	3	3	5
	Chills	3	2	12
	Headache	6	7	7
	Pain	1	2
	Hypotension	1	3	5
	Fatigue	1	1
	Blurred vision	1	1
	Hypoxia	1		1
	Sore throat	1	1
Total			19	13	1

All patients developed mild CRS with grades 1–3 fever, chills, headache, nausea (9/9 patients), and occasional grades 1–2 hypotension (3/9 patients). The most common grades 2 and 3 toxicities (55% each) were fever and chills, starting postinfusion and lasting up to 72 hours, followed by headache and hypotension. Hypotension was mild and did not require vasopressors. Hypoxia was rare (one patient had one episode) and did not require high-flow oxygen. Six out of nine patients developed grades 1–2 CRS with grade 1 episodes after 16 of 75 infusions and grade 2 CRS episodes after 10 of 75 infusions. Retrospective CRS grading according to ASTCT criteria did not change initial CTCAE V.5.0 grades. Nausea with or without emesis occurred in eight of nine patients (any grade). Only three (33%) patients developed grade 2 lower extremity pain that was managed with ibuprofen and lasted for up to 3 days. No DLTs were observed, and the MTD was not reached. The phase II cohort experienced similar toxicities, and there were no DLTs (table 3). All toxicities resolved to less than grade 3 within 72 hours.

Clinical responses

Phase I component

In phase I patients, clinical activity of GD2 BATs was limited to ungraded responses, with no objective responses according to INRC or RECIST. Six of nine patients (3/5 NB, 2/3 OST, 1/1 DSRCT) had PD, and three patients (2/5 NB, 1/3 OST) had SD. Six patients (one OST and five NB) received additional therapy after completing the study. Five of nine patients (one OST, four NB) survived 1-year post-GD2BAT therapy. The median OS was 18.0 months for phase I group (figure 1B).

Although phase I patients IT20111 and IT20115 did not have objective responses, their clinical courses were noteworthy. Patient IT20111, with NB infiltrating the BM alongside progressive metastatic bone and soft tissue disease, underwent treatment at dose level 1. Remarkably, GD2BAT therapy yielded a complete BM response (figure 2A) and stable disease in the other metastatic sites for approximately 6 months. Six months post-therapy, the patient resumed topotecan and cyclophosphamide chemotherapy, the same regimen this patient had previously failed, and achieved an MIBG response and improvement of limping and pain. The disease progressed 36 months after GD2BAT therapy, leading to the patient’s demise 47 months after treatment.

Figure 2.

(A) A bone marrow response with clearance of neuron-specific enolase (NSE) positive tumor cells in the bone marrow of an NB patient IT20111 after 8 infusions of GD2BATs. (B) The metabolic PET changes after five infusions of GD2 BAT in an OST patient IT20115. Arrow marks point the intracranial frontal and right maxillary sinus tumor extention before and after GD2BATs infusions. (C) Flow cytometry staining for CD4+ and CD8+ to evaluate presence of tumor infiltrating lymphocytes in the tumor for Pt # 20 115. Most intranasal tumor infiltrating lymphocytes were CD8⁺. Cells were gated on CD3⁺ cells and analyzed for FITC positive CD4 and PE positive CD8 T cells. (D) The proportion of CD3+A1 G4+ cells in the peripheral blood mononuclear cells of patient # 8 preinfusion and at various times during and after GD2BATs infusions, % of CD3⁺A1G4⁺ cells shows the persistence of GD2BATs in peripheral blood. CD3⁺A1G4⁺ cells were measured by using two color FACS gated on lymphocytes using anti-CD3 and A1 G4 anti-idiotypic antibody for hu3F8. PET, positron emission tomography; SUV, standardized uptake value.

In dose level 2, patient IT20115 with recurrent radiation-induced nasopharyngeal OST exhibited an early metabolic PET response with a significant reduction in ¹⁸FDG PET uptake after the sixth GD2BAT infusion (figure 2B). A biopsy of the soft tissue nasal mass revealed substantial infiltration of CD8+ cells (figure 2C). The patient attained stable disease by RECIST criteria and was initiated on pazopanib and remained stable allowing for tumor resection, which extended his remission for a year. Unfortunately, the disease progressed 14 months after GD2BAT therapy.

Phase II component

In the phase II group comprising 10 evaluable patients with NB, 1 patient (IT00013) achieved a PR according to INRC criteria, 5 patients had SD, and 4 patients had PD. Notably, additional clinical activity was observed in some patients with overall SD, including a patient with soft tissue mass response and another a BM response. The median OS for the phase II group was 31.2 months. The patient with PR was monitored without additional therapies and progressed 6 months later. All five patients with SD were initiated on additional therapies.

Highlighted phase II cases

Patient IT00013 with recurrent NB and multiple soft tissue thoracic and abdominal masses experienced a PR in all lesions after GD2BAT infusions (figure 3A), accompanied by improvement in MIBG scan. Patient IT00031 had PD per INRC criteria based on new skeletal MIBG uptake but concurrently achieved a PR in the dural intracranial soft tissue mass (figure 3B). Patient IT00033 achieved SD after GD2BAT therapy, with a stable MIBG scan. Subsequent restart of prior salvage chemotherapy (topotecan and cyclophosphamide), led to further improvement, resulting in a negative MIBG scan 1 year after GD2BAT. The patient is alive with disease 4 years after GD2BAT. Patient IT 000040 achieved an overall INRC SD but cleared BM disease at the same time.

Figure 3.

(A) Patient’s IT00013 retroperitoneal soft tissue mass (white arrow mark) before (left panel) and after (right panel) eight infusions of GD2BATs. (B) Patient’s IT00031 intracranial dural mass (yellow arrow mark) before (left panel) and after (right panel) eight infusions of GD2BATs.

Immune evaluations

Persistence of GD2BATs

Staining for IgG2a (OKT3 component of the BiAb) was inconclusive due to high background staining. Subsequently, PBMCs were stained with anti-idiotypic antibody A1G4 specific for hu3F8 in one patient (IT00005) at MSK. Increasing numbers of circulating GD2BATs were detected above baseline (figure 2D) after GD2BAT infusions and up to 2.4% of PBMCs were GD2BATs detected in the circulation 6 days after the last infusion.

Anti-GD2 cytotoxic activity

Specific anti-GD2 cytotoxicity mediated by PBMC, measured through IFN-γ EliSpots 1 month after therapy, exhibited a significant increase compared with pretherapy levels in the combined 10 phase II NB patients in response to KCNR (figure 4A, p<0.02), and MG63 (figure 4B, p<0.02) cell lines, while responses to K562 were not statistically significant. Conversely, in the eight phase I patients, post-therapy IFN-γ EliSpots responses to KCNR, MG63 and K562 (NK target) did not show significant differences. When IFN-γ EliSpots responses to KCNR and MG63 by PBMC of the total 18 phase I and II patients were analyzed collectively, post-therapy IFN-γ EliSpots were significantly higher than pretherapy levels for KCNR (p<0.03) and MG63 (p<0.04). Changes in NK activity (figure 4C,F) was not evident in either phase I or II patients (figure 4C,F). Notably, patient IT 20111 exhibited a remarkable cellular immune response correlated with a complete BM response and overall SD PBMCs from the patient demonstrated 7, 4.1, and 8.8-fold increases in IFN-γ ELISpot responses to KCNR, MG63, and K562 cell line stimulation, respectively.

Figure 4.

Anti-GD2 cytotoxicity (IFN-γ EliSpots). Each panel shows the pre-IT and post-IT IFN-γ EliSpots responses from PBMC after overnight stimulation. Phase II portion: (A–C) Responses of PBMC from the phase II NB patients to KCNR (p<0.02), MG63 (p<0.02), and K562 (NS). Phase I/II combined: (D–F) IFN-γ EliSpots responses of PBMC from all 18 of the phase I/II patients pre-IT and post-IT after overnight exposure to KCNR (p<0.03), MG83 (p<0.04), and K562 (not significant, NS), respectively. PBMC, peripheral blood mononuclear cells.

Serum cytokine/chemokine responses

Cytokines associated with immune responses and CRS were assessed tested pre and post-IT in 18 patients from the combined phase I/II cohort. The results demonstrated a significant increase in post-IT levels of IL-12 (p<0.002) compared with pre-IT levels (figure 5A). Similarly, in the phase II group of 10 patients, post-IT levels of IL-12 were significantly elevated (figure 5B) over pre-IT levels (p<0.02). Post-IT serum levels of MIP1β (figure 5C) and IL-10 (figure 5D) from the combined 18 phase I/II patients were significantly higher than pre-IT levels (p<0.02 for both). Notably, post-IT levels of IL-6, TNFα, and IP-10 in 18 phase I/II patients did not show significant elevation compared with pre-IT levels (data not shown). No changes were observed in pre-IT or post-IT results for IFN-γ EliSpot responses of PBMC to KCNR, MG63, and K562, and no significant alterations were noted in serum cytokine levels for IL-12, TNF-α, IL-10, IL-6, MIP1β, IP-10 in 8 phase I patients (data not shown). Additionally, there were no significant changes between pre-IT and post-IT levels of TNFα, MIP1β, and IP-10 in the phase II patients (data not shown).

Figure 5.

Cytokines/chemokines: (A) (IL-12) shows significant changes between pre-IT and post-IT levels of IL-12 in all 18 phase I/II patients (p<0.002). (B) (IL-12) shows a significant change between pre-IT and post-IT IL-12 levels for 10 NB in the phase II patients analyzed separately (p<0.02). (C) (MIP-1β) shows post-IT levels of MIP-1β increased significantly over pre-IT levels in 28 phase I/II patients (p<0.02). (D) (IL-10) shows the levels of IL-10 significantly increased from pre-IT to post-IT in all 18 phase I/II patients (p<0.02).

Discussion

Disialoganglioside GD2 serves as a mediator of cell adhesion, potentially contributing to increased invasion and motility of cancer cells27 and exhibits dense expression on NB cells.2 These properties make GD2 an effective target for immunotherapy.28

Despite the administration of anti-GD2 mAbs, a substantial proportion (10%–20%) of high-risk NB patients fail to achieve remission, and 50%–60% of those receiving immunotherapy post-SCT experience relapse.3 4 Following unsuccessful initial treatment and chemoimmunotherapy, subsequent therapeutic modalities, including MIBG, kinase inhibitors,29 anti-GD2 antibodies,30 31 vaccines, and haploidentical SCT32 have not induced high response rates in recurrent or refractory NB. Various anti-GD2 cell-based therapies have been employed,33 including several reported trials involving GD2 CAR-T cells31 32 34 35 and one trial with GD2 CAR NKT cells.36 The primary distinctions between CAR-based therapies and BATs are as follows: In most cases, CAR-T therapies involve the creation of viral genetic constructs transduced into T effector cells, whereas the BiAb preparation used to arm ATC is produced by heterconjugation of clinical grade mAbs. CAR-Ts need to encounter their targets (in vivo tumor cells) to become activated, while BATs are activated on expansion in vitro and subsequently reactivated in vivo by engagement with tumor. CAR-Ts are long-lasting cells capable of persisting and expanding significantly in vivo in response to stimulation, presenting both tumor control potential and the risk of significant complications, including CRS. In contrast, BATs are preapoptotic cells with limited lifespan that undergo limited expansion after infusions. Notably, clinical trials of CAR T cells targeting GD2+ have struggled to demonstrate significant objective responses,37 38 with one recent exception reported by an Italian group demonstrating a 63% ORR.39 OST accounts for a significant portion of cancer related deaths (8.9%) in the pediatric age group,28 yet there have been no new drugs for OST in over three decades.40 The combined progression-free survival from seven phase II trials in OST patients with measurable disease was modest (12% at 4 months), with less than 5% objective responses observed in multiple over the past two decades.41

Our extensive experience with BATs targeting HER2+, EGFR+, CD20+ in phase I/II trials for metastatic breast cancer,13 14 metastatic hormone refractory prostate cancer,15 16 and metastatic pancreatic cancer13 17–19 22 demonstrated an absence of DLTs in over 180 patients. Encouraged by these results, we applied this platform to safely target GD2+ tumors.

It was feasible to collect PBMCs by apheresis, expand the T cells, harvest, and arm ATC with GD2Bi, aliquot and cryopreserve multiple doses, and infuse the GD2BATs in pediatric outpatient clinics at MSK, CHM, and UV. GD2BAT therapy was well tolerated in this highly pretreated pediatric patient population. Notably, visceral pain commonly associated with anti-GD2 mAb infusions was not a problem, and we observed mild leg pain in only two patients. The MTD was not reached, and infusions were associated with mild CRS symptoms, with the most frequent side effects being chills, fever, headache, fatigue, and hypotension, all resolving within 72 hours and not exceeding grade 2 toxicity. The study, conducted before the ASTCT definition for CRS, was published, used CTCAE V.5.0 for scoring, with retrospective grading according to ASTCT criteria24 showing consistency between CTCAE and ASTCT grading, especially low-grade CRS. All CRS cases were either CTCAE/ASTCT grade 1 or grade 2, primarily featuring fever and hypotension that responded well to fluids.

The postinfusion percentage of GD2BATs in PBMCs, as assessed by staining for the hu3F8 component of GD2BATs, was elevated up to 15% (see figure 4D). Subsequent doses showed a decline in postinfusion levels of circulating BATs, likely attributed to the elevation of overall PBMC counts due to the continuous administration of GM-CSF and IL-2. It is noteworthy that some BATs remained detectable in circulation even at 6 days postinfusion.

Despite the absence of objective radiological responses in the phase I study, noteworthy clinical activity was observed in two patients with NB and OST. Patient IT20111 clinical response correlated with elevated anti-NB CTL and NK activity, as well as increased levels of Th₁ cytokines/chemokines. In the phase II portion, several patients exhibited clinical activity, and immunologic sensitization may have positively affected responses to subsequent therapies when patients responded to chemotherapy regimens that they failed before IT. These observations suggest that GD2BAT infusions may “immunosensitize” tumors, rendering them more responsive to subsequent chemotherapy—a phenomenon we also observed in metastatic pancreatic cancer.13 17

More than half of the patients developed specific cellular anti-GD2 CTLs, with elevated anti-GD2 endogenous CTL activity observed in both phases I and II patients. This clinical study supports the notion that multiple BAT infusions up to 1.6×10⁹ BATs/kg induce endogenous, specific antitumor CTL responses, akin to our observations in breast, prostate, and pancreatic cancer patients.13 14 16 Importantly, this may “sensitize” tumors, allowing previously ineffective chemotherapy to regain efficacy when reintroduced. Such immunosensitization is consistent with cases where patients became responsive to previously ineffective chemotherapy regimens after immunotherapy with checkpoint inhibitors.42

While no statistically significant correlations between the phenotypic composition of cell products and clinical responses were found, all but one patient’s GD2BATs contained at least 30% of CD8+ cells. Patients whose GD2BATs contained low proportions of CD8 T cells exhibited low cytotoxicity against anti-GD2+ tumor cells. For example, the GD2BATs of patient 20119 containing only 6% CD8 cells exhibited 3% cytotoxicity against the GD2+ tumor cells and no subsequent clinical tumor response was noted. Furthermore, GD2 BATs from patients displaying BM responses or PR in soft tissue lesions exhibited high levels in vitro killing of GD2+ tumor lines. (IT20111–53%, IT00013–47.9%, IT00040–43.1%; average cytotoxicity was 29% for all patients).

This phase I/II study demonstrates that IT with GD2BATs is safe and feasible in heavily pretreated and resistant/refractory patients who have failed previous therapy, including GD2 mAb therapies. The extension cohort of 10 patients in the phase II portion not only confirmed safety but also revealed a median OS of 31 months in these heavily pretreated patients which exceeds the reported median OS of 16.1 months among 372 patients experiencing their first recurrence of NB who underwent various second-line treatments5 including anti-GD2 mAbs, and to the reported median OS of 16.1 months in 71 patients with recurrent/refractory disease treated with chemotherapy-only regimens.6 Additionally, it contrasts with the median OS of 2.1 months observed in 97 patients who exclusively received palliative treatment following their initial first relapse.5 These results provide a robust rationale for conducting a phase II trial to assess the clinical benefits of GD2BATs in patients with refractory/resistant NB. Since the cell quantity targeted for production in dose level 3 of our study approached the maximum technically feasible cell numbers that we could achieve and noteworthy clinical activity was observed through all dose levels, we propose to use the dose level 2 of 80 million GD2BATs/kg/infusion in the future expanded phase 2 trial.

Data availability statement

Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by Wayne State University, Phase I Institutional Review Board, # 1403012875, Memorial Sloan Kettering Cancer Center, Institutional Review Board # 15-206A(4), University of Virginia Institutional Review Board for Health Sciences Research # 19031. Participants gave informed consent to participate in the study before taking part.