Treatment and outcome of adult-onset neuroblastoma

Abstract

Adult-onset of neuroblastoma is rare and little is known about its biology and clinical course. There is no established therapy for adult neuroblastoma. Anti-GD2 immunotherapy is now standard therapy in children with high-risk neuroblastoma; however, its use has not been reported in adults. Forty-four adults (18–71 years old) diagnosed with neuroblastoma between 1979 and 2015 were treated at Memorial Sloan Kettering Cancer Center. Five, 1, 5, and 33 patients had INSS stage 1, 2, 3, and 4 diseases respectively. Genetic abnormalities included somatic ATRX (58%) and ALK mutations (42%) but not MYCN-amplification. In the 11 patients with locoregional disease, 10-year progression-free (PFS) and overall survival (OS) was 35.4±16.1% and 61.4±15.3% respectively. Among 33 adults with stage 4 neuroblastoma, 7 (21%) achieved complete response (CR) after induction chemotherapy and/or surgery. Seven patients with primary refractory neuroblastoma (all with osteomedullary but no soft tissue disease) received anti-GD2 antibodies, mouse or humanized 3F8. Antibody-related adverse events were similar to those in children, response rate being 71.4%. In patients with stage 4 disease at diagnosis, five-year PFS was 9.7± 5.3% and most patients who were alive with disease at 5 years died of neuroblastoma over the next 5 years, 10-year OS being only 19.0±8.2%. Patients who achieved CR after induction had superior PFS and OS (p=0.006, p=0.031 respectively). Adult-onset neuroblastoma appeared to have different biology from pediatric or adolescent NB, and poorer outcome. Complete disease control appeared to improve long-term survival. Anti-GD2 immunotherapy was well tolerated and might be beneficial.

Introduction

Neuroblastoma (NB), an embryonal cancer arising from sympathetic nervous system is the most common extracranial solid tumor in children totalling 600–700 new cases per year in the United States. The median age of diagnosis is 17 months¹ and >90% of patients are diagnosed before 5 years of age.² Very rarely, NB is diagnosed in adults ; little is known about the natural history of the disease in this age group. Reports on adult-onset NB are often reviews of large NB databases such as the Surveillance Epidemiology and End Results (SEER) in the United States or the International NB Research Group (INRG). The former database identified 144 patients ≥20 years old at diagnosis representing 6.1% of all patients (n=2054) diagnosed with NB between 1973 and 2002³ and 35 patients ≥60 years old at diagnosis representing 0.9% patients diagnosed between 1973 and 2010.⁴ The INRG database covering patients diagnosed between 1990 and 2002, identified 200 (2.3%) patients between the ages of 10 and 21 years but outcome analysis combined all patients >10 years. ⁵ Other reports include esthesioneuroblastoma, a disease with a very different natural history, e.g. a report on 118 patients included 86 patients with head and neck primaries likely to be esthesioneuroblastoma. ⁶ Finally most prior reports, including one from our group ⁷, combine adults with adolescents, assuming that the natural history and outcomes are similar in these two groups. Table 1 summarizes salient publications on adult-onset NB.

Table 1.

Summary of previous reports on adults with neuroblastoma:

Authors	Years of data collection	No. of pts ≥18 years old	Age (range)	Number of pts with non-head and neck primaries	Information on genetics	The number of pts with no disease progression to the date of publication*	Median follow up for PFS (range)* (months)	The number of patients who survived to the date of publication*	Median follow up for OS (range)* (months)
Tang et al.8	1948–1972	15	N.R.	N.R.	N.R.	N.R.	N.R.	N.R.	N.R.
Mackay et al.9	N.R.	9	18–72	5	N.R.	3	12 (10–24)	3	16 (12–24)
Dosik et al.10	1974–1976	3	20–37	3	N.R.	2	8 (1–18)	2	9 (2–19)
Rowe et al.11	N.R.	2	26–29	2	N.R.	1	26 (10–36)	1	34 (10–57)
Lopez et al.12	1950–1977	3	33–52	3	N.R.	2	22 (9–23)	2	22 (16–23)
Aleshire. et al.13	1947–1983	8	26–75	3	N.R.	1	12 (7–32)	1	12 (7–32)
Kaye et al.14	1983–1984	3	20–38	2	N.R.	1	9 (6–12)	1	11 (6–16)
Allan et al.15	1977–1981	3	25–34	3	N.R.	0	16 (5–30)	0	16 (5–30)
Prestidge et al.16	1967–1987	5	18–34	5	N.R.	N.R.	N.R.	2	44 (10–133)
Franks et al.17	1968–1995	10	18–36	10	N.R.	1	24 (4–83)	3	43 (15–197)
Kushner et al.7	1985–2001	18	18–41	18	0/18 MYCNA	1	24 (8–130)	8	45 (11–160)
Esiashvili et al. (SEER)3	1973–2002	125	≥20	N.R.	N.R.	N.R.	N.R.	N.R.	3yr-45.9%, 5yr-36.3%**
Podda et al.18	1982–2001	12	18–69	8	0/3 MYCNA	1	26 (6–108)	1	50 (15–124)
Mosse et al. (INRG)5	1990–2002	N.R.	N.R	N.R.	N.R.	N.R.	N.R.	N.R.	N.R.
Berbegall et al.19	1997–2012	7	18–60	7	1/7 MYCNA	0	3 (1–11)	4	1 (1–3)
Jrebi et al.20	1980–2009	15	19–66	15	N.R.	2	N.R.	5	72 (3–480)
Conter et al.6	1994–2012	118	18–82	32	N.R.	N.R.	N.R.	N.R.	3-yr: L1 94%, L2 83%, M 68% 5-yr: L1 90%, L2 73%, M 33% 10-yr: L1 69%, L2 41%, M 13% ***
Rogowitz et al.4	1973–2007	216	≥18	N.R.	N.R.	N.R.	N.R.	N.R.	N.R.
Mazzocco et al.21	1979–2009	4	18–27	4	0/4 MYCNA	0	33 (5–87)	2	70 (33–130)

^*Patients with head and neck primaries were excluded;

^**Primary sites in these patients were not reported;

^***Patients with head and neck primaries were included. The survival rate was assessed for each stage by INRG classification.

Abbreviations:

het: hetero, INRG: International Neuroblastoma Risk Group, MYCNA: MYCN amplification, No.: number, N.R.: not reported, pts: patients, OS: overall survival, PFS: progression free survival, SEER: Surveillance, Epidemiology, and End Results, yr:year

When adolescent and adult-onset NBs were combined as a group and compared to NBs arising before 10 years of age, clear biological differences were noted: somatic MYCN amplification, which confers an aggressive phenotype, was reported in 9% of adolescents and adults⁵ whereas the incidence in children is 20–25 %. ^{5, 22} MYCN-amplified tumors have almost never been observed in adult-onset NB, with only one report in the literature of a patient ≥18 years at diagnosis. ¹⁹ Conversely, loss-of-function alpha thalassemia/mental retardation syndrome X linked (ATRX) mutations and deletions are more commonly observed in NB arising in patients ≥12 years old compared to the younger age group. ²³ Although mechanisms of oncogenesis of ATRX mutations/deletions are unclear, they could be related to slower growth and chemoresistant phenotype of adolescent/adult NB.

Adolescents and adults with NB are reported to have a worse outcome than children ^{3, 5, 17, 18}; however, due to the rarity of the disease, reports on the former group have included small numbers of adults. The SEER review reported 3 and 5-year overall survival (OS) rates of 45.9% and 36.3% respectively in adults ≥20 years at diagnosis. Corresponding rates in adolescents (10–19 years) were 61.3% and 46.2% respectively. ³ Similar results were noted in an INRG review with 5-year OS of 46% in patients diagnosed above age 10 years. ⁵

Whereas a multimodality regimen including chemotherapy, surgery, radiation therapy and antibody-mediated anti-GD2 immunotherapy has now been established as standard therapy for high-risk NB in children,^24–28 no accepted approach exists for adults, who in general, cannot tolerate high-dose chemotherapy as well as children. Furthermore, the recently-approved anti-GD2 antibody dinutuximab, which has been shown to improve survival,^{27, 28} is only indicated for pediatric use, and little information exists on its use in adults.

We previously published the clinical outcome and therapeutic details on the largest series of adolescents and adults (≥12 years) treated at a single institution.⁷ We now focus solely on the clinical course, therapy, and outcomes of NB diagnosed in adulthood (≥18 years at diagnosis), include data on an additional 26 adults and an updated follow up on the previously reported patients.

Methods

Patients

After obtaining permission from the institutional review board of Memorial Sloan Kettering Cancer Center (MSKCC), medical records of all patients diagnosed with NB above the age of 18 years were retrospectively reviewed. Diagnosis of NB was made either by confirmation of histopathology by Department of Pathology at MSKCC or the demonstration of bone marrow (BM) involvement with extrinsic cells plus elevated urinary catecholamines. All included patients had completed staging investigations at diagnosis. Staging was performed using International Neuroblastoma Staging System (INSS).²⁹ Response was assessed by International Neuroblastoma Response Criteria (INRC); complete response (CR): no evidence of tumors, very good partial response (VGPR): decrease of primary tumor by 90–99% and negative metastatic tumors, partial response (PR): > 50% reduction of all measurable tumors and bone disease with no more than 1 positive BM site, mixed response (MR): >50% reduction of any measurable lesion with <50% reduction in any other, no response (NR): <50% reduction of any measurable lesion but <25% increase in any existing lesion, progressive disease (PD): any new lesion or increase of any measurable lesions by >25%.²⁹ Tumor genetic mutations were tested either by whole genome sequencing,²³ DNA sequencing for anaplastic lymphoma kinase (ALK) gene, or exome sequencing with a comprehensive genomic profile focused on cancer-related genes using Foundation One (Foundation Medicine)³⁰ or Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT).³¹

Statistical methods

Progression-free survival (PFS) and OS were assessed from the time of diagnosis (SPSS software, IBM). The Kaplan-Meier method was used to estimate the survival probabilities. Log-rank test was used to test univariate associations between survival (PFS, OS) and the following variables: gender, age, stage, year of diagnosis, biological characteristics, and response to the induction chemotherapy and/or surgery. Variables that had significance on univariate analysis (p<0.05) were assessed by multivariable analysis using a Cox Regression model.

Results

Patient demographics

Forty-four patients (25 males and 19 females) with adult-onset NB diagnosed between 1979 and 2015 were included. Median age at diagnosis was 25 (range 18 to 71) years and median follow up was 5.3 (range 1.0 to 37.4) years. 17 patients were diagnosed and received their initial therapy at MSKCC.

Clinical and Radiological Features at diagnosis

Common presenting symptoms at diagnosis were back pain (n=15, 34%), abdominal pain (n=13, 30%), and leg pain (n=5, 11%). In two patients (5%), NB was found incidentally by routine physical examination or imaging tests done after trauma. Two patients (5%) were hypertensive at diagnosis. Urine catecholamines were elevated in 20/28 (71%) patients tested. The most common primary site was adrenal (n=26, 59%) followed by abdominal paraspinal region (n=11, 25%), and mediastinum (n=5, 11%). Two patients (5%) had only bone and BM disease without an obvious primary site. Imaging studies to detect metastases included: ^99mTc-methylene diphosphonate bone scan alone (n=15; 34%), Iodine123 (¹²³I)-metaiodobenzylguanidine (MIBG) scan alone (n=9; 20%), fluorodeoxyglucose-positron emission tomography (FDG-PET) alone (n=4; 9%), and combinations of the above scans including ¹³¹I-MIBG scan (n=3) in the remaining 16 (34%) patients. BM aspiration and biopsy was performed in 36 patients and was positive in 18 (50%). Eleven patients (stage 1: n=5, stage 2: n=1, stage 3: n=5) had locoregional disease, and 33 patients had metastatic stage 4 NB. Of the 33 stage 4 patients, 4 did not have osteomedullary disease but only distant soft tissue disease (i.e. had stage 4N NB). Metastatic sites in patients with stage 4 disease included bone and/or BM (n=29), cervical (n=4), and mediastinal soft tissue (n=3). Seven patients had been misdiagnosed at diagnosis, one each with Wilms tumor, desmoplastic small round cell tumor and primitive neuroectodermal tumor, and two each with pheochromocytoma and lymphoma, prior to their NB diagnosis.

Biological Features

On histology, 34 patients were deemed to have NB and 10 to have ganglioneuroblastoma (GNB). One tumor had elements of pheochromocytoma in addition to NB. No patients (all tested by fluorescence in situ hybridization; n=40) had MYCN-amplified NB. However, somatic ATRX mutations or deletions were detected in 15/26 (58%) tumors tested and somatic ALK mutations were detected in 10/16 (42%). Eleven patients underwent broader targeted exome sequencing at initial diagnosis (n=3) or at relapse (n=8) by Foundation One and/or MSK-IMPACT. These tumors harbored ATRX (n=3) or ALK mutations (n=2). However, an increase in number and diversity of mutations was noted at relapse. (Table 2) One patient had serial exome sequencing at first and second relapse. While the initial sample had CSF1R mutations, the sample collected at second relapse 40 months later had acquired 2 additional mutations.

Table 2.

Results of targeted exome sequencing in 11 adult patients

Pt #	Timing of sample	Test	Mutation noted
1	Diagnosis	Foundation One	ALK R1275Q, ATRX Y1808fs*12, EIF4E3-FOXP1 fusion
2	Diagnosis	Foundation One	ATRX loss, BRCA1 Q1756fs*74, IKZF2 R477*, NF2 Q410*
3	Diagnosis	MSK-IMPACT	ALK F1174L, ATRX G2088V
4	Relapse	Foundation One	ALK F1174L, MUTYH G382D
5	Relapse	Foundation One	ATRX deletion exon 6–9, ALK D1276_R1279>E
6	Relapse	Foundation One	NF1 splice site 7971-2A>T, TP53 V157I
7	Relapse (1st)	Foundation One	CSF1R V32G
	Relapse (2nd)	MSK-IMPACT	CRLF2 deletion (Xp22.33), FAT1 deletion (4q35.2), NUP93 S654G
8	Relapse	MSK-IMPACT	ATRX R2164T, BLM R786S, PTPRD intragenic deletion (9p24.1), RAC1 rearrangement (c.35+2000_345+45dup)
9	Relapse	MSK-IMPACT	EZH2 splice site 2029+1del23, SMARCA4 splice site 3081+2T>C, TP53 N235S,
10	Relapse	MSK-IMPACT	ERBB3 L441M, NOTCH3 SP1264RT, PAK1 S155_N157del, RBM10 G22V
11	Relapse	MSK-IMPACT	ATRX W2077*, BCL6 V614L, KIT D52del, RASA1 P924S

Treatment for locoregional NB

Of the 11 patients with locoregional NB, 8 received induction chemotherapy and all had surgery. Nine patients achieved CR, 1 achieved PR, and 1 patient had PD. (Table 3) After induction therapy and/or surgery, 3 patients underwent autologous stem cell transplant (ASCT); there was 1 ASCT-related mortality.

Table 3.

Response to induction therapy

Stage	Overall INRC response	Soft tissue response	Bone disease	Bone marrow disease
1 (N=5)	5 CR (100%)	5 CR (100%)	NA
2 (N=1)	1 PR (100%)	1 PR (100%)	NA
3 (N=5)	4 CR (80%) 1 PD (20%)	4 CR (80%) 1 PD (20%)	NA
4 (N=33)	7 CR (21%) 7 PR (21%) 12 NR (36%) 7 PD (21%)	14 CR (45%*) 14 NR (45%*) 3 PD (10%*)	4 CR (14%**) 9 PR (31%**) 9 NR (31%**) 6 PD (21%**)	6 CR (33%***) 12 NR (67%***)

^*2 patients did not have soft tissue disease at diagnosis;

^**4 patients did not have bone disease at diagnosis and 1 patient had evaluation of bone disease after induction;

^***BM disease was observed in 18 patients at diagnosis

Abbreviations:

CR: compete response, NA: not applicable, NR: no response, PD: progressive disease, PR: partial response, INRC: the International Neuroblastoma Response Criteria

Treatment for stage 4 NB

Initial Therapy

Of the 33 patients with stage 4 NB, 27 received high-dose induction chemotherapy containing cyclophosphamide, doxorubicin, vincristine, cisplatin, and etoposide. One patient, aged 71 years at diagnosis initially received crizotinib for ALK-mutated NB. The remaining 5 did not receive high-dose chemotherapy because they had been misdiagnosed and received therapy for their respective initial diagnoses prior to the corrected diagnosis of NB. Surgery was attempted in 11 patients. Surgery was not attempted in the remaining 22 patients: 2 had no obvious primary tumor, 7 had PD after induction, 12 were considered to have unresectable disease, and surgery was deemed to be too risky in the 71-year old patient. There was no induction-related mortality. Overall response after induction chemotherapy and/or surgery was 7 CR, 7 PR, 12 NR, and 7 PD. Responses were noted both in soft tissue and osteomedullary disease. (Table 3)

Post-induction Therapy

Of the 7 stage 4 patients achieving CR, 5 underwent ASCT, 1 underwent ASCT followed by anti-GD2 immunotherapy with murine 3F8 (m3F8), and 1 did not receive any post-induction therapy. Seven patients with less than complete responses (all with osteomedullary but no soft tissue disease) after induction chemotherapy received anti-GD2 immunotherapy: 6 with m3F8 with (n=4) or without (n=2) granulocyte macrophage colony-stimulating factor (GM-CSF), and one with humanized 3F8 (hu3F8) without GM-CSF. Among those receiving anti-GD2 immunotherapy, 1 patient received ¹³¹I-m3F8 as consolidation before m3F8.²⁶ Adults experienced similar adverse events with m3F8 and hu3F8 as previously described in children: pain, allergic reactions and blood pressure changes.³² Most received treatment as outpatients with m3F8/hu3F8 being administered over 30–90 minutes. Response rate after anti-GD2 monoclonal antibody (mAb), m3F8 or hu3F8, in patients with primary refractory disease was 71.4% (5CR, 2 NR). External beam radiotherapy to the primary site (18–37Gy) was administered to 22 patients. One patient (71 years old at diagnosis) with somatic ALK mutation was initially treated with crizotinib and had an objective response not amounting to PR. He had PD after 11 months and was switched to ceritinib on which he continues with stable disease for 26+ months. A further patient also received crizotinib for primary refractory osteomedullary without any objective response.

Patterns of relapse and survival

Fourteen of 44 patients are alive with a median survival of 89.0 months. Six (14%) patients did not experience relapse. For patients with locoregional NB, 5-year PFS and OS was 70.7±14.3% and 81.8±11.6% respectively. However, 4 patients relapsed >5 years after diagnosis, resulting in 10-year PFS and OS of 35.4±16.1% and 61.4±15.3% respectively. Sites of relapse in patients with locoregional NB were localized at the primary site alone (n=4) and primary soft tissue with osteomedullary disease (n=3). For patients with stage 4 disease at diagnosis, 31/33 (94%) patients had relapse or disease progression with 5-year PFS and OS of 9.7±5.3% and 58.8±8.8 respectively. However, 11 patients died of NB >5 years after diagnosis yielding 10-year OS of 19.0±8.2. (Figure 1) Sites of progression were soft tissue alone (n=8), bone/BM alone (n=18) or both (n=12). Treatment at relapse included chemotherapy (high-dose: n=14, low to intermediate dose: n=8), surgery (n=5), radiotherapy (n=10), MIBG therapy (n=4) and anti-GD2 immunotherapy (n=8). Eight patients, including 1 in second CR were treated with m3F8 (n=6) or hu3F8-containing (n=2) regimens. Treatment was well tolerated with adverse events similar to those receiving immunotherapy after induction. The patient treated in second CR remained in remission for 63 months. Of the 7 patients treated with evaluable disease, one achieved CR, 5 had SD, and 2 had PD yielding a response rate of 12.5%.

Figure 1.

Patients with locoregional NB had superior PFS and OS than those who had metastatic disease at diagnosis in univariate analysis.

Prognostic factors for survival

When all 44 patients were considered by univariate analysis, patients < 25 years old at diagnosis (n=22; p=0.007), those with locoregional NB (n=11; p=0.005), and those who achieved CR after induction and/or surgery (n=16; p<0.001) had significantly superior PFS. Those with locoregional NB (p=0.023) and those who achieved CR after induction and/or surgery (p=0.002) had superior OS. (Table 4A) Diagnosis before the median year of diagnosis, 2000 (n=23) was not prognostic for survival (p=0.975 for PFS, p=0.136 for OS). In multivariate analysis, only achieving CR post induction was independently associated with favorable PFS (95% confidence interval (CI) 0.151 – 0.789) and OS (95% CI 0.140 – 0.842). (Table 4B) Analysis of only the 33 stage 4 patients showed that CR after induction was the only prognostic factor for PFS and OS (n=7; p=0.006 and p=0.031 respectively). (Table 4A, Figure 2)

Table 4A.

Univariate analysis for Survival

Variables			p value for PFS	p value for OS
All patients (N=44)
gender	male (n=25)	female (n=19)	0.563	0599
age	<median (25years) (n=22)	≥ median (25years) (n=22)	0.007*	0.095
year of diagnosis	<median 2000 (n=23)	≥ median 2000 (n=21)	0.975	0.136
stage	stage 1–3 (n=11)	stage 4 (n=22)	0.005*	0.023*
response after the induction and/or surgery	CR (n=16)	refractory (n=28)	<0.001*	0.002*
Stage 4 NB (N=33)
gender	male (n=20)	female (n=13)	0.865	0.354
age	<median (26years) (n=17)	≥ median (26years) (n=16)	0.283	0.292
year of diagnosis	<median 1999 (n=13)	≥ median 1999 (n=20)	0.684	0.283
Osteomedullary disease	positive (n=29)	negative (n=4)	0.335	0.726
BM histology (28 pts tested)	positive (n=18)	negative (n=10)	0.311	0.649
response after the induction and/or surgery	CR (n=7)	refractory (n=26)	0.006*	0.031*

Table 4B.

Multivariate analysis for Survival in all patients (N=44)

Variables	p value	Hazard ratio	95% confidence interval
For PFS
age <median (25years)	0.061	0.509	0.251 – 1.032
stage 1–3	0.105	0.440	0.161 – 1.186
CR after the induction and/or surgery	0.012*	0.345	0.151 – 0.789*
For OS
stage 1–3	0.186	0.482	0.163 – 1.422
CR after the induction and/or surgery	0.020*	0.343	0.140 – 0.842*

Figure 2.

Achieving CR after induction and/or surgery was associated with favorable PFS and OS in adults with stage 4 NB.

Discussion

Adult-onset NB is a rare entity and this data represents the largest single-institution report detailing clinical management and outcomes. Among the adults in our study, most patients were diagnosed with NB before the age of 40, similar to that reported in SEER review,³ although the oldest patient was diagnosed at age 71 years. Sites of primary tumors and metastatic disease resembled those seen in children with NB. Presenting symptoms were typically non-specific, with back pain and abdominal pain being the most common. Persistence of these symptoms especially in young adults should be followed by appropriate scans in an attempt to make an appropriate diagnosis. The incidental discovery of NB was relatively rare occurring in only 5% of our patients. Unfortunately, 16% of adults in our study were misdiagnosed and incorrectly treated at diagnosis (all at institutions outside MSKCC) and received suboptimal/incorrect initial treatment. Small round blue cell tumors in adults are rare, and malignancies such as lymphoma and Ewing’s family of tumors should be considered. A knowledge of the immunohistochemical staining (synaptophysin, chromogranin for NB ³³) and genomic changes can yield the correct diagnosis and initiation of appropriate therapy. As compared to children, a higher proportion of adults presented with metastatic disease at diagnosis: 75% of patients presenting with stage 4 disease, whereas in children, only 40–55% have stage 4 disease at diagnosis. ^34–37 This might be related to the delay in diagnosis of adult-onset NB; a longer interval between the onset of symptoms and diagnosis has been reported in adults.¹⁸ However, in our study, 64% of patients with locoregional disease at diagnosis went on to have metastatic disease at relapse.

Pediatric solid tumors in general, and NB in particular have a low frequency of mutations³⁸, but genetic testing on tumors in adults yielded a significantly different pattern of genomic aberrations. Similar to other reports on adult-onset NB ^{7, 18, 19, 21}, none of our patients had MYCN-amplified NB. (Table 1) This is in contrast to pediatric NB where 20–25% of patients have MYCN-amplified NB ^{5, 22}. However, other somatic aberrations particularly those in ALK and ATRX genes were far more common: ALK mutations were noted in 42% and ATRX mutations/deletions in 58% compared to reported incidence of 8–10% ^{39, 40} and 4–12 % ^{23, 41} respectively in children. In fact, all 11 tumors undergoing exome sequencing had at least one genomic aberration. The very low incidence of MYCN-amplification and common finding of ATRX mutation suggest that the drivers of tumorigenesis in adult-onset NB differ from those in children. ATRX is thought to play an essential role in maintaining chromosome stability by interacting with histones at telomeres and by regulating cell cycles. ^{19, 42} Adult malignancies, in general, have a much higher rate of somatic mutations which are thought to accumulate over time.⁴³ Similar to other adult-onset tumors, the development of NB could be related to the accumulation of somatic mutations with age.

In our cohort, 61% patients started therapy outside MSKCC, and therapies varied significantly over time (patients studied were diagnosed over a period of 36 years). Hence outcomes cannot necessarily be correlated with administered treatment. However, some general conclusions can be drawn. PFS and OS were both inferior in adults than in those reported in children. Though the number of adults with non-metastatic disease at diagnosis in our cohort was small (n=11), their outcome (10-year PFS of 35%) appeared to be inferior to children with non-metastatic disease in whom PFS is typically >80% ^{44, 45} in the absence of MYCN-amplification. In addition, adults with locoregional disease were prone to late relapses; 4 of 7 patients experiencing relapse did so beyond 5 years from diagnosis. For adults with stage 4 disease, induction and consolidation therapy was variable due to the long time period over which patients were studied, because many patients received part or most of their therapy at institutions outside MSKCC and because there is no acceptable standard therapy for adults with NB. Achieving CR after induction chemotherapy and/or surgery was the only factor that impacted PFS and OS in adults suggesting that an aggressive approach to achieving remission is warranted in this cohort, while balancing chemotherapy-related toxicities. The positive impact of complete surgical resection on patients with locoregional NB in adults has been reported by other investigators.²⁰ We had previously observed that adolescents with NB tend to have an indolent course, however, in the adults with metastatic disease this does not appear to be the case: 5-year PFS was only 9.7% and most patients who were alive with disease at 5 years died of NB over the next 5 years, 10-year OS being only 19.0%. Similar to children or adolescents, the first relapse/progression occurred early: median PFS after diagnosis being 27±6 months in adults. However, adolescents had longer post-relapse survival ⁵ unlike adults, who had rapid progression after relapse. Possible reasons for this difference in post-relapse survival between adolescents and adults could be the salvage therapies administered. Children and adolescents are generally treated with more aggressive salvage regimens e.g. we used high-dose cyclophosphamide-based salvage chemotherapy for refractory or relapsed NB in children at MSKCC with major responses being observed in 50% of adolescents.⁷ However, not all adult patients can tolerate high-dose chemotherapy. In our adult cohort, only 14/31 adults with relapsed metastatic NB received high-dose salvage chemotherapy. Less toxic therapies including intermediate/low-dose chemotherapy and radiotherapy were administered to the majority of patients.

We previously reported on the anti-NB effects of anti-GD2 mAb, m3F8 and hu3F8 on minimal residual or chemorefractory osteomedullary disease.^{46, 47} A similar therapeutic effect was observed in adults with a 71.4% response in primary refractory disease. Adults were able to tolerate hu3F8 or m3F8 therapy well, mostly in the outpatient setting with a toxicity profile similar to that seen in children: pain and allergic reactions being the most common adverse events. We were unable to formally assess the association between anti-GD2 mAb therapy and survival because of the limited number of patients who had CR or PR. While only 16 patients were deemed eligible for anti-GD2 immunotherapy at the end of induction and/or surgery, the responses observed in patients with osteomedullary disease are encouraging. With the regulatory approval of dinutuximab and dinutuximab-beta, and the clinical development of hu3F8, anti-GD2 immunotherapy is more widely available and should be incorporated into multimodality therapy for all adults with high-risk NB. Because adult-onset NB probably develops over years, the acquisition of multistep somatic mutations might make these tumors more targetable by T-cells. Thus T-cell mediated immunotherapy including immune checkpoint inhibition, which has not hitherto been effective in pediatric solid tumors ^{48, 49} could potentially be more useful in adults, though clinical data on the use of such agents is currently unavailable.

Somatic genomic aberrations appeared to be more common in adults with NB. Hence we particularly recommend exome or whole genome sequencing in adult-onset NB to detect actionable genomic aberrations. Indeed, we were able to successfully target ALK mutations in one patient with the US Food and Drug Administration (FDA)-approved ALK inhibitors; although the response was modest, this 71-year old patient, who was not a suitable candidate for chemotherapy, tolerated therapy well for over 3 years. Unfortunately, the loss-of-function ATRX mutations are not currently targetable, though modulation of methylation might have a therapeutic effect. In addition to somatic genomic aberrations, germline mutations and polymorphisms may be important for assessing the different in tumorigenesis between pediatric and adult NB. ALK and paired-like homeobox 2b (PHOX2B) germline mutations are associated with familial NB,⁵⁰ and several polymorphism including LMO1, CASC15, LIN28B can be associated with increased susceptibility of NB. ⁵¹ Germline genomic testing and genome-wide association studies for adult patients are likely to shed further light on the pathophysiology of this rare group of patients. Germline testing in 4 patients in our cohort revealed aberrations in two. (one with BRCA1 mutation, the other with TP53 and NF1 mutations).

The review of the SEER database previously showed that the outcome of adult-onset NB improved over three decades from 1970s to 1990s ³, however, our study did not show any improvement in outcome between before and after 2000. The SEER report also mentioned a substantial decline in incidence of adult-onset NB over those 3 decades; it is possible that some patients with misdiagnoses could have been included in earlier data.³ Updated SEER data on the outcome of recently diagnosed adult-onset NB will show the trend of the outcome over time.

In conclusion, adult-onset NB appears to have different biology with higher frequency of ALK mutation or ATRX mutation/deletion and no MYCN amplification from pediatric NB. Furthermore, adults with NB appear to have worse outcomes regardless of stage at diagnosis compared to pediatric or adolescent patients, likely related to the difficulty in diagnosing and treating NB because of lack of information on adult-onset NB. Complete disease control appears to improve long-term survival. Anti-GD2 immunotherapy, m3F8 or hu3F8, is tolerated well in adults and might be beneficial. Because adult patients tend to have high relapse rate regardless of stage at diagnosis and few options for salvage therapy, maintenance therapy after induction and surgery may be important to prevent relapse. Given the success of immunotherapy in pediatric NB and the higher frequency of somatic mutations in adults, there is reason to examine the benefits of both immunotherapy and targeted small molecules in adult NB. A prospective multicenter study that investigates the genome-wide analyis and newer therapeutic interventions including and anti-GD2 immunotherapy for adults with neuroblastoma is warranted.

Novelty and Impact.

Adult-onset neuroblastoma is extremely rare. Little is known about its biology and clinical course, and standard treatment has not been established. Anti-GD2 immunotherapy is now incorporated into multi-modal therapy in children with high-risk neuroblastoma. We describe here the clinical course, genomic features, prognostic factors, and treatment including anti-GD2 immunotherapy in the largest cohort of adults with neuroblastoma.

Abbreviations

ALK

anaplastic lymphoma kinase

ASCT

autologous stem cell transplant

ATRX

alpha thalassemia/mental retardation syndrome X linked

BM

bone marrow

CI

confidence interval

CR

complete response

FDA

Food and Drug Administration

FDG-PET

fluorodeoxyglucose-positron emission tomography

GNB

ganglioneuroblastoma

GM-CSF

granulocyte macrophage colony-stimulating factor

GWAS

genome-wide association studies

Hu3F8

humanized 3F8

I

iodine

INRC

International Neuroblastoma Response Criteria

INRG

International Neuroblastoma Risk Group

INSS

International Neuroblastoma Staging System

MAb

monoclonal antibody

MIBG

metaiodobenzylguanidine

MR

mixed response

MSKCC

Memorial Sloan Kettering Cancer Center

MSK-IMPACT

Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets

M3F8

mouse 3F8

NB

neuroblastoma

NA

not applicable

No.

number

NR

no response

N.R.

not reported

OS

overall survival

PD

progressive disease

PFS

progression-free survival

PHOX2B

paired-like homeobox 2b

PR

partial response

Pts

patients

SEER

Surveillance, Epidemiology, and End Results

VGPR

very good partial response

Yr

year