Anti-GD2 Antibody Dinutuximab Beta and Low-Dose Interleukin 2 After Haploidentical Stem-Cell Transplantation in Patients With Relapsed Neuroblastoma: A Multicenter, Phase I/II Trial

PURPOSE Patients with relapsed high-risk neuroblastoma (rHR-NB) have a poor prognosis. We hypothesized that graft-versus-neuroblastoma effects could be elicited by transplantation of haploidentical stem cells (haplo-SCT) exploiting cytotoxic functions of natural killer cells and their activation by the anti-GD2 antibody dinutuximab beta (DB). This phase I/II trial assessed safety, feasibility, and outcomes of immunotherapy with DB plus subcutaneous interleukin-2 (scIL2) after haplo-SCT in patients with rHR-NB. METHODS Patients age 1-21 years underwent T-/B-cell–depleted haplo-SCT followed by DB and scIL2. The primary end point ‘success of treatment’ encompassed patients receiving six cycles, being alive 180 days after end of trial treatment without progressive disease, unacceptable toxicity, acute graft-versus-host-disease (GvHD) ≥grade 3, or extensive chronic GvHD. RESULTS Seventy patients were screened, and 68 were eligible for immunotherapy. Median number of DB cycles was 6 (range, 1-9). Median number of scIL2 cycles was 3 (1−6). The primary end point was met by 37 patients (54.4%). Median observation time was 7.8 years. Five-year event-free survival (EFS) and overall survival from start of trial treatment were 43% (95% CI, 31 to 55) and 53% (95% CI, 41 to 65), respectively. Five-year EFS among patients in complete remission (CR; 52%; 95% CI, 31 to 69) or partial remission (44%; 95% CI, 27 to 60) before immunotherapy were significantly better compared with patients with nonresponse/mixed response/progressive disease (13%; 95% CI, 1 to 42; P = .026). Overall response rate in 43 patients with evidence of disease after haplo-SCT was 51% (22 patients), with 15 achieving CR (35%). Two patients developed GvHD grade 2 and 3 each. No unexpected adverse events occurred. CONCLUSION DB therapy after haplo-SCT in patients with rHR-NB is feasible, with low risk of inducing GvHD, and results in long-term remissions likely attributable to increased antineuroblastoma activity by donor-derived effector cells.


INTRODUCTION
Patients with high-risk neuroblastoma (HR-NB) have 5-year survival rates of approximately 50%, whereas patients with metastatic disease at relapse show a 4-year progression-free survival of 6% and overall survival (OS) of 15%. 1,2 More recently, the combination of anti-GD2 immunotherapy and chemotherapy (chemoimmunotherapy) showed promising overall response rates (ORRs) in patients with relapsed HR-NB (rHR-NB). 3,4 The anti-GD2 antibody dinutuximab beta (ch14.18/CHO; DB) is approved as frontline postconsolidation therapy in HR-NB. 5,6 DB acts through antibody-dependent cell-mediated cytotoxicity (ADCC) and complementdependent cytotoxicity (CDC). 7,8 Previous cytotoxic therapies may impair the ability of natural killer (NK) cells to mediate ADCC. 9 Therefore, reconstitution of functional NK cells by transplantation of stem cells from haploidentical family donors (haplo-SCT) before immunotherapy is an appealing concept, as NK cells Trial treatment. From day 60 after transplantation, patients without GvHD or acute GvHD (aGVHD) #grade 2 were scheduled to receive DB as an 8-hour infusion of 20 mg/m 2 once per day on five consecutive days, for a total of six cycles given every 4 weeks. To avoid induction of GvHD, low-dose scIL2 was added only in cycles 4-6 on days 6, 8, 10 (1 3 10 6 IU/m 2 ; Fig 1). Patients exhibiting complete response (CR), partial response (PR), or stable disease (SD) after cycle three received three more cycles. In case of response after cycle 6, patients were eligible to receive another three cycles. Following protocol recommendations, continuous morphine infusions were routinely administered during DB treatment.
Immunosuppressive medication had to be stopped before DB treatment. Chemotherapy, experimental anticancer medication, and radiotherapy were not allowed during immunotherapy.
The study Protocol was approved by the appropriate authorities and institutional review boards. All legal guardians and/or patients provided written informed consent before screening. This trial was registered with ClinicalTrials.gov (identifier: NCT02258815) and EudraCT (2009-015936-14).

Study Assessments
Standard tumor response evaluation was not part of the study protocol but was expected before haplo-SCT, as previous observations suggested remission status as a major factor for outcomes. 13 Response evaluation using 1993 International Neuroblastoma Response Criteria before DB treatment, after cycles 3, 6, and 9 (if applicable), after 1 year, and annually thereafter was mandatory. 14 Evaluations included mIBG scintigraphy (International Society of Pediatric Oncology, European Neuroblastoma [SIOPEN] mIBG score 15 ), bone marrow aspirates, and whole-body magnetic resonance imaging or magnetic resonance imaging-computed tomography scans of tumor sites, according to the RECIST. 16 Bone marrow (BM) samples were analyzed according to Mehes et al and later published international guidance, including CONTEXT Key Objective Survival rates for patients with relapsed high-risk neuroblastoma are poor. This study examined feasibility, safety, and response to an immunotherapeutic regimen of dinutuximab beta and low-dose subcutaneous interleukin-2 after haploidentical stem-cell transplantation (haplo-SCT) in patients with relapsed high-risk neuroblastoma.

Knowledge Generated
Five-year event-free survival and overall survival from start of trial treatment were 43% and 53%, respectively. Overall response rate and complete response rate in 43 patients with evidence of disease after haplo-SCT were 51% and 35%, respectively. Toxicity profile and treatment-related mortality of the combinational treatment were favorable with a low frequency of graft-versus-host disease. Relevance (S. Bhatia) Immunotherapy with dinutuximab beta after haplo-SCT is feasible, safe, and results in long-term remissions. These findings inform the next steps that include definitive randomized trials to determine the role of the individual components of the therapeutic regimens.* *Relevance section written by JCO Associate Editor Smita Bhatia, MD, MPH. microscopy and minimal disease (MD) evaluation with automatic immunofluorescence detection of GD2-/CD56positive neuroblastoma cells. [17][18][19] All MIBG scans were submitted to independent central review. For details on response criteria, see appendix 3. Toxicity was recorded according to Common Terminology Criteria for Adverse Events (CTCAE4.0).

Statistical Analysis
The primary end point 'success of treatment' was defined as patients receiving six cycles of DB, alive 180 days after end of trial treatment, without progression, and unacceptable toxicity or acute GvHD $ grade 3 or extensive chronic GvHD according to Glucksberg or Seattle classification, respectively. 20,21 Treatment success of $50% was considered relevant with a minimum of 35 evaluable patients for assessing efficacy with a Simon's two-stage design (significance level 5%; power 80%), 22   before DB treatment, after cycles 3 and 6, after 1 year, and annually thereafter were integral part of this trial. Evaluations included mIBG scintigraphy, bone marrow aspirates (including minimal disease measurement with AIPF), and whole-body MRI or MRI-CT scans of tumor sites. AIPF, automatic immunofluorescence detection system; CT, computed tomography; DB, dinutuximab beta; haplo-SCT, haploidentical stem-cell transplantation; IL2, interleukin 2; mIBG therapy, 131I-meta-iodobenzylguanidine therapy; MRI, magnetic resonance imaging; s.c., subcutaneous.  11.1% of cycles. Lower dosages and prolonged infusion rates followed trial recommendations and were instigated as a result of hypersensitivity reactions. Twenty-nine patients (42.6%) did not complete trial treatment (six cycles): 13 patients (19.1%) because of PD, six because of therapyrelated toxicity (hypersensitivity/inflammatory reactions), four because of hemolytic anemia, two because of posterior reversible encephalopathy syndrome (PRES)/central nervous system toxicity, and one each because of human herpes virus 6 (HHV-6) infection and bacterial sepsis; two patients decided to stop immunotherapy. Of the 29 patients discontinuing treatment, 15 received 1-2 cycles and 14 received 3-5 cycles of DB. Twenty-one patients received more than six cycles: three patients received seven cycles, and 18 received nine cycles. In 62 patients (91.2%), scIL2 was administered as prescribed; in five patients (7.3%), administration was unknown. One patient (1.5%) did not receive scIL-2 because of hypersensitivity reactions, but continued with DB treatment.
Toxicity. Treatment-related adverse events (AEs) are summarized in Table 2. Hematologic grade 3/4 AEs occurred in 29 patients (42.6%), with hemolytic anemia reported in six patients (8.8%). Most nonhematologic Abbreviations: CR, complete remission; DB, dinutuximab beta; EFS, event-free survival; haplo-SCT, haploidentical stem-cell transplantation; mIBG, metaiodobenzylguanidine; MD, minimal disease; MR, mixed response; NR, nonremission; OS, overall survival; PD, progressive disease; PR, partial remission; SCT, stem-cell transplantation; SD, stable disease. a EFS and OS were calculated from start of trial treatment (first antibody cycle in this trial, ie, first day of first DB cycle, after haplo-SCT). b, c Patients were grouped according to the median times from first relapse to haploidentical stem-cell transplantation (291 days) and start of DB treatment (415 days). grade 3/4 AEs were fever, pain, hypersensitivity reactions, capillary leak syndrome, elevated liver enzymes, and central neurotoxicity. Sixty-two patients (91.2%) experienced pain in cycle 1, which decreased to 26 patients (38.2%) by cycle 6. Anaphylactic/inflammatory reactions requiring intensive care treatment were observed in six patients (8.8%). Viral, fungal, and bacterial infections occurred in five patients, three of whom died: one because of HHV-6 infection with encephalitis/pneumonitis and two because of bacterial infections. Severe peripheral neurotoxicity with transient paresthesia occurred in one patient. One patient died after the second DB cycle with signs of encephalitis and/or PRES. This patient had a tumor infiltrating skull and dura, pre-existing absence epilepsy, and Opsoclonus-Myoclonus-Ataxia Syndrome.
Occurrence of late-onset aGvHD during DB treatment was low (n 5 5; 7.4%). Two patients developed grade 2 and 3 aGvHD of the gut. Three patients (4.4%) developed grade 1/2 skin GvHD, without the need for systemic therapy. No grade 4 GvHD was reported. No additional AEs were observed in the presence of low-dose scIL2.

DISCUSSION
Treatment of rHR-NB remains challenging with poor survival rates. 2,24 Here, we investigated the feasibility, safety, and outcome of DB in combination with low-dose scIL2 after haplo-SCT in a cohort of patients with rHR-NB.
The use of ex vivo T-cell-depleted haplo-SCT takes advantage of high-dose chemotherapy and NK-mediated alloreactive graft-versus-tumor/leukemia GvL effects. In mismatched SCT, NK-mediated GvL effects reduced the relapse rates in patients with leukemia. 10,25,26 We previously showed that haplo-SCT was associated with low incidence of GvHD and TRM in patients with rHR-NB. 13 However, the 5-year EFS of 19% indicated insufficient graft-versus-tumor effects. 13 It has been shown that ADCC can augment posttransplant antitumor activity of donor-derived effector cells. 27,28 Since DB improves outcomes after autologous SCT during first-line treatment, 5,29 administration of DB may also augment graft-versus-neuroblastoma effects after haplo-SCT through early expanding and persisting donorderived NK cells . 30 Previously, we reported 5-year EFS and OS of 19% and 23%, respectively, in patients with rHR-NB receiving haplo-SCT without antibody treatment, whereas the current trial exhibits 5-year EFS and OS rates of 43% and 53%, respectively. 13 Disease status before immunotherapy and before haplo-SCT influenced prognosis. Patients with CR or PR before immunotherapy had significantly better 5-year EFS and OS than those with NR/MR/PD. Tumor responses were observed in patients with macroscopic residual disease before DB, as demonstrated by an ORR of 51.2%. Of the 43 patients with disease after haplo-SCT, approximately half reduced their tumor load with DB treatment and 35% achieved CR. This is likely due to an interplay between DB, donorderived effector cells, and CDC. 30 Similar results were reported in the HR-NBL1/SIOPEN trial investigating DB after ASCT in frontline therapy 29 ; however, it is uncertain whether the autologous immune system can still exert antineuroblastoma activity in the relapse setting after intensive chemotherapy. A limitation of this study is that it only represents the proportion of rHR-HB patients without initial rapid progression during individual relapse treatments.  However, on the basis of previously reported relapse trials, our cohort appears to be a comparable collective in terms of risk factors (eg, number of relapses, time to relapse, and MYCN amplification status). 2,3,13 The results of trials evaluating the role of combinational treatment of anti-GD2 antibodies with chemotherapy in rHR-NB have recently reported comparable ORRs. 3,4 These approaches avoid potential side effects of HSCT, especially GvHD, whereas in our approach, donor-derived effector cells in combination with antibody treatment could provide a stronger, longer-lasting tumor control. Because of the different designs, shorter observation times, and subsequent therapies in several patients in the chemoimmunotherapy trials, a direct comparison with our results is currently limited. A randomized trial would be necessary to demonstrate the superiority of one approach. A combination of both approaches, reinduction with (B) Patients receiving full protocol treatment (per-protocol population), that is, response represents the remission status after completion of trial treatment, defined as six full cycles of DB treatment. Dropout patients are listed in the upper part of B (PD before treatment completion and dropout side effects or AE). Percentages in the lower part of B refer to patients treated according to protocol (n 5 39) and to the whole cohort of 68 patients (intention-to-treat population). CR maintained: patients who started DB treatment in CR; CR improved: patients who achieved CR during/after DB treatment. Event-free survival and overall survival were calculated from start of trial treatment (first antibody cycle in this trial, ie, first day of first DB cycle after haplo-SCT). AE, adverse event; CR, complete remission; DB, dinutuximab beta; EFS, event-free survival; haplo-SCT, haploidentical stem-cell transplantation; MR, mixed response; NR, nonremission; OS, overall survival; PD, progressive disease; PR, partial remission; SD, stable disease. (E) EFS: BM infiltration before first DB cycle; (F) OS: BM infiltration before first DB cycle; (G) EFS: age at study entry; and (H) OS: age at study entry. Event-free survival and overall survival were calculated from start of trial treatment (first antibody cycle in (continued on following page) chemoimmunotherapy and consolidation with haplo-SCT followed by DB, could be another option. The majority of our patients were DB-naïve; 10 patients received anti-GD2 therapy during relapse or frontline treatment. OS, EFS, and ORR did not differ significantly between DB-naïve or previously exposed patients. Since repeated anti-GD2 antibody exposure in the relapse setting has been shown to be effective, we assume that this also applies to our approach. 3 We performed two additional retrospective analyses evaluating factors at diagnosis, at relapse, and those related to haplo-SCT. None of the factors at diagnosis maintained independent prognostic value, likely attributable to the relatively small sample size. One significant factor was 131 I-mIBG therapy before conditioning given in some patients with residual mIBG-avid disease. Interestingly, independent of remission status before haplo-SCT, 131 I-mIBG therapy had a significantly positive influence on OS and EFS in univariate and MVA. In vitro data suggest that radiation can induce immunogenic tumor cell death and release of tumor-specific antigens, NK cell ligands, and stress-inducible proteins, which could be identified and attacked by the donor-derived immune system and NK cells cotransfused during haplo-SCT. 31,32 Remission , PR before haplo-SCT as well as before DB treatment was another factor with independent prognostic value. In univariate analysis, female sex was associated with worse OS, which was not confirmed in MVA.
The toxicity of haplo-SCT was acceptable with low TRM and aGvHD rates (7.5%), which was lower than that reported after allogeneic SCT with matched donors. 33 Only two patients developed aGvHD grade 2 and 3 each during DB treatment. These cases can either be considered as late-onset aGvHD (.100 days post-transplant), or de novo GvHD induced by the antibody treatment. In both patients, DB was continued after resolution of GvHD without recurrence. Hemolytic anemia is a well-known complication after allogeneic SCT with an incidence of about 6%, 34 we cannot exclude that the antibody treatment might have induced or aggravated the hemolysis seen in 9% of our patients. The most frequently reported grade 3/4 AEs observed here were similar to those reported during DB treatment after ASCT. 29 Three patients died during DB therapy because of HHV-6 infection, bacterial infection, and PRES. The infections were considered to be associated with intensive pretreatment and SCT. The encephalopathy might have occurred due to a combination of disease-, transplant-, and antibodyrelated toxicities. Relapse remained the major cause of death with a CI of 49% at 5 years.
For further optimization, we suggest replacing the CD3/CD19 depletion by T-cell receptor ab/CD19-depleted grafts, which results in accelerated immune reconstitution and allows cotransfusion of additional gdT cells with potential antitumor and antiviral activity. 35 Toxicity might be reduced by using the 10-day continuous infusion for DB as previously reported. 36 The use of checkpoint inhibitors to optimize the efficacy of DB is also currently being explored. 37 The additional use of scIL2 is questionable since our trial was not designed to evaluate specific effects of scIL2. 30 SIOPEN data showed that adding scIL2 to DB in the frontline setting does not improve efficacy but increases toxicity. 29 Thus, we believe that the addition of IL-2 should not be considered in future trials. In summary, DB therapy after haplo-SCT demonstrated antitumor activity with acceptable toxicity in patients with rHR-NB and was associated with notable EFS and OS among patients who had achieved at least PR with previous therapy.
Further prospective and randomized trials are warranted to evaluate the contribution of each component of the approach, and larger cohorts are needed to allow better risk stratification and patient selection. Abbreviations: CR, complete remission; DB, dinutuximab beta; EFS, event-free survival; haplo-SCT, haploidentical stem-cell transplantation; HR, hazard ratio; mIBG, meta-iodobenzylguanidine; MR, mixed response; MVA, multivariate analysis; OS, overall survival; PD, progressive disease; PR, partial remission; SD, stable disease. a EFS and OS were calculated from start of trial treatment (first antibody cycle in this trial, ie, first day of first DB, cycle after haplo-SCT). b Global test for testing if a variable is significant in the multivariate model.