Identification of a minimal region of loss on chromosome 6q27 associated with poor survival of high-risk neuroblastoma patients

ABSTRACT

Patients with high-risk neuroblastoma (HR-NB) often initially respond to therapy, but afterward they become resistant and disease recurred. Unfortunately, it does not exist one or more specific chromosome defects associated with relapse or refractory NB. Recently, genomic evidence from primary tumors indicated that the distal region of chromosome 6q is loss in HR-NB patients with fatal outcome. We identified a minimal common region of loss of chromosome 6q27 spanning an area of 2.09 Mb by high-resolution DNA copy number data of a small cohort of HR-NB samples carrying 6q loss. This region of loss harbored five genes T, SFT2D1, RPS6KA2, FGFR1OP, and UNC93A. We found that low SFT2D1, RPS6KA2, and FGFR1OP gene expression predicted poor outcome in HR-NB patients using public R2 Platform. Further functional studies will be essential to confirm the presumed tumor suppressor gene(s) located within 6q27 region. These results suggest that SFT2D1, RPS6KA2, and FGFR1OP genes may be responsible for poor prognosis of HR-NB tumors with 6q27 loss, and their haploinsufficiency may be crucial in accelerating tumor progression.

KEYWORDS:

• Neuroblastoma
• distal 6q27 loss
• tumor suppressor genes
• smallest region of overlap
• haploinsufficiency

Introduction

Neuroblastoma (NB), arising from neural crest progenitor cells, is the most common tumor in very young children.¹ Although unbelievable improvements have been made for children with low-risk disease, patients with high-risk neuroblastoma continue to have an incurable disease with survival rates <40% in spite of cytotoxic chemo-radiotherapy.^2,3 Patients with high-risk neuroblastoma (HR-NB) initially respond to therapy, but afterward up to 60% of these become resistant and disease recurred.^4-7 NB is one example of how specific chromosome rearrangements can influence the trend of the disease. Currently, whole-genome analysis revealed large genomic DNA copy number alterations (losses of chromosome 1p, 3p, 4p, 11q and gains of 1q, 2p, 17q) that are of prognostic significance for newly diagnosed NB patients.^8-15 It has been proposed that multiple genes located on recurrent chromosome segmental aberrations may have dosage effects that influence tumor formation.^14,15 Although mutational events are rare and often non-recurrent, combining these findings with chromosome copy number aberrations and pathway analysis it has been demonstrated the recurrent involvement of genes implicated in neuritogenesis and chromatin remodeling.^16-19 Unfortunately, it does not exist one or more specific chromosome defects associated with relapse or refractory NB. Previous studies have delineated a 7 Mb region at distal 6q27 loss in HR-NB patients with extremely poor survival outcome.²⁰ A complete genomic analysis of the distal 6q region would help to identify the putative NB tumor suppressor gene(s).

Here, we report the cytogenetic characterization of 15 primary HR-NB samples with 6q loss. We determined the size and genetic content of these losses by high-resolution array-Comparative Genome Hybridization (a-CGH), and decoded a common region of overlap spanning 2.09 Mb. We hypothesize that the genes contained in the commonly loss region are dosage-sensitive, and that the haploinsufficiency of one or more of these genes was associated with extremely poor survival outcome of HR-NB.

Results

Analysis of 6q loss in HR-NB patients

Array-CGH disclosed multiple segmental chromosome gains or losses in the genomic profile of 15 NB samples analyzed. Partial chromosome losses were most often detected for chromosome arms 1p, 3p, 4p, 11q, and partial chromosome gain for 1q, 2p, 7q, 17q abnormalities associated with grim survival outcome of NB patients.^8-15 All NB samples harbored 6q loss: seven cases showed distal loss, six had interstitial loss, and two carried both distal and interstitial 6q losses. Figure 1 gives an overview of all rearrangements in chromosome arm 6q. We remarked that only three samples with 6q loss also showed MYCN amplification (Table 1).

Table 1. Genome copy number analysis of 15 NB samples that define a minimal common region of loss on chromosome arm 6q.

Case number	Age at diagnosis (months)	Stage	MYCN status	Immunotherapy (anti-GD2)	Relapse	Overall survival	Genomic positions and sizes of 6q loss (NCBI Build 37)
NB01	44	Stage 4	Non-amplified	Yes	Yes	Dead of disease	Chr 6q: start 109862108 – end 170892907 Cytoband: 6q21-q27 Size: 61.03 Mb
NB02	36	Stage 4	Amplified	No	No	Dead of disease	Chr 6q: start 84548022 – end 97711237 Cytoband: 6q14.2-q16.1 Size: 13.16 Mb
NB03	73	Stage 4	Non-amplified	No	Yes	Dead of disease	Chr 6q: start 77442663 – end 87862867 Cytoband: 6q14.1-q14.3 Size: 10.42 Mb
NB04	68	Stage 4	Non-amplified	No	Yes	Dead of disease	Chr 6q: start 154726548 – end 170591649 Cytoband: 6q25.2-q27 Size: 15.86 Mb
NB05	46	Stage 4	Non-amplified	No	Yes	Dead of disease	Chr 6q: start 150435942 – end 170892302 Cytoband: 6q25.1-q27 Size: 20.45 Mb
NB06	45	Stage 4	Non-amplified	Yes	Yes	Dead of disease	Chr 6q: start 61971892 – end 170892907 Cytoband: 6q11.1-q27 Size: 108.92 Mb
NB07	43	Stage 4	Non-amplified	Yes	No	Dead of disease	Chr 6q: start 65112507 – end 170892907 Cytoband: 6q12-q27 Size: 105.78 Mb
NB08	92	Stage 4	Non-amplified	No	Yes	Dead of disease	Chr 6q: start 76312616 – end 79735696 Cytoband: 6q14.1 Size: 3.42 Mb and Chr 6q: start 123888421 – end 170890108 Cytoband: 6q22.31-q27 Size: 47 Mb
NB09	42	Stage 4	Non-amplified	No	No	Dead of disease	Chr 6q: start 148743936 – end 167927956 Cytoband: 6q24.3-q27 Size: 19.18 Mb
NB10	51	Stage 4	Non-amplified	No	No	Dead of disease	Chr 6q: start 76631274 – end 131520645 Cytoband: 6q14.1-q23.2 Size: 54.88 Mb and Chr 6q: start 138865808 – end 170890108 Cytoband: 6q23.3-q27 Size: 32.02 Mb
NB11	23	Stage 4	Non-amplified	Yes	No	Dead of disease	Chr 6q: start 120464984 – end 170892302 Cytoband: 6q22.31-q27 Size: 50.42 Mb
NB12	71	Stage 4	Non-amplified	No	Yes	Dead of disease	Chr 6q: start 166203071 – end 170890108 Cytoband: 6q27 Size: 4.68 Mb
NB13	7	Stage 4	Amplified	Yes	Yes	Dead of disease	Chr 6q: start 109577950 – end 167536510 Cytoband: 6q21-q27 Size: 57.95 Mb
NB14	20	Stage 4	Amplified	Yes	Yes	Dead of disease	Chr 6: start 136268631 – end 144860512 Cytoband: 6q23.3-q24.2 Size: 8.59 Mb
NB15	74	Stage 4	Non-amplified	Yes	No	Dead of disease	Chr 6q: start 70805204 – end 99914626 Cytoband: 6q13-q16.2 Size: 29.10 Mb

Figure 1. Schematic representation of chromosome arm 6q rearrangement. Distribution of breakpoints of chromosome 6q in 15 HR-NB samples. Green lines on the right side of the ideogram represent losses; the red line represents gain. The indicated five genes in the 6q27 region meet the following criteria: statistically significantly lower expression of the gene in NB cell lines with vs without 6q loss, at least one mutation in the gene as described in primary NB, and statistically significantly worse survival outcome when gene expression is low in HR-NB. SRO = smallest region of overlap.

Of interest, we defined a minimal common region of genomic loss in distal 6q by a sample that showed a small loss spanning an area of 4.68 Mb. The 6q27 region was deleted from oligomer A_14_P103168 (166.203.071 bp) (first deleted) to oligomer A_14_P120743 (170.890.108 bp) (last deleted), containing 38 genes (C6orf176, LOC441177, T, PRR18, SFT2D1, BRP44L, RPS6KA2, MIR1913, RNASET2, FGFR1OP, CCR6, GPR31, TCP10L2, UNC93A, TTLL2, TCP10, C6orf123, C6orf124, MLLT4, HGC6.3, KIF25, FRMD1, DACT2, SMOC2, THBS2, WDR27, C6orf120, PHF10, TCTE3, C6orf70, C6orf122, C6orf208, LOC154449, DLL1, FAM120B, PSMB1, TBP, and PDCD2). Among the cases presenting distal 6q loss one sample showed a gain adjacent to the 6q loss. In particular, this sample carried in 6q27 region a focal gain from oligomer A_14_P200779 (168303033 bp) (first gained) to oligomer A_14_P120743 (170890108 bp) (last gained), containing 20 genes (MLLT4, HGC6.3, KIF25, FRMD1, DACT2, SMOC2, THBS2, WDR27, C6orf120, PHF10, TCTE3, C6orf70, C6orf122, C6orf208, LOC154449, DLL1, FAM120B, PSMB1, TBP, and PDCD2). So by subtraction we can define a critical Smallest Region of Overlap (SRO) of 2.09 Mb containing five genes T, SFT2D1, RPS6KA2, FGFR1OP, and UNC93A that met at least one of following standards: (i) decrease of expression in NB cell lines carrying 6q loss,²⁰ (ii) loss of function mutations reported in primary HR-NB tumors,¹⁶ and (iii) low expression associated with worse overall HR-NB patient survival.²⁰

Neuroblastoma patient outcomes based on the expression of the genes contained in the critical SRO in distal 6q27 loss

Next, we evaluated the association of the expression of the genes contained in the critical SRO in distal 6q loss (T, SFT2D1, RP56KA2, FGFR1OP, and UNC93A genes) with HR-NB patient outcomes. We obtained online microarray analysis results of two independent RNA-sequencing datasets (SEQC n = 498 tumors; TARGET-Asgharzadeh n = 249 tumors) with available survival endpoints. T gene probe-sets (SEQC probe-set NM_003181; TARGET-Asgharzadeh probe-set 2984500), SFT2D1 gene probe-sets (SEQC probe-set NM_145169; TARGET-Asgharzadeh probe-set 2984573), RPS6KA2 gene probe-sets (SEQC probe-set NM_021135; TARGET-Asgharzadeh probe-set 2984655), FGFR1OP gene probe-sets (SEQC probe-set NM_007045; TARGET-Asgharzadeh probe-set 2936564), and UNC93A gene probe-sets (SEQC probe-set NM_018974; TARGET-Asgharzadeh probe-set 2936731) were chosen for the highest average signals for analysis. The most appropriate cutoff for survival analyses was established in the expression value where the separation of survival curves reached the maximum (log-rank test).

These analyses revealed that low expression of SFT2D1, RPS6KA2, and FGFR1OP genes was associated with worse event-free and overall survival of HR-NB patients consistently across the two datasets (Figures 2 and 3). In contrast, high expression of T and UNC93A genes was significantly associated with reduced event-free and overall survival especially in HR-NB patients (Figures 4 and 5).

Figure 2. Neuroblastoma high-risk patient outcomes based on SFT2D1, RPS6KA2, and FGFR1OP gene expression using the neuroblastoma SEQC-498 patient datasets. Kaplan–Meier analyses and comparison of SFT2D1, RPS6KA2, and FGFR1OP expression with outcome. HR-NB patients were divided into high (blue) and low (red) gene expression groups, with patient numbers in parentheses. (a) Event-free survival (Expression cutoff: 7.931, min.grp = 8), and (b) overall survival (Expression cutoff: 7.840, min.grp = 8) curves for SFT2D1 gene expression are shown. (c) Event-free survival (Expression cutoff: 25.386, min.grp = 8), and (d) overall survival (Expression cutoff: 25.386, min.grp = 8) curves for RPS6KA2 gene expression are shown. (e) Event-free survival (Expression cutoff: 17.599, min.grp = 8), and (f) overall survival (Expression cutoff: 17.484, min.grp = 8) curves for FGFR1OP gene expression are shown.

Figure 3. Neuroblastoma high-risk patient outcomes based on SFT2D1, RPS6KA2, and FGFR1OP gene expression using the neuroblastoma TARGET-Asgharzadeh-249 patient datasets. Kaplan–Meier analyses and comparison of SFT2D1, RPS6KA2, and FGFR1OP expression with outcome. HR-NB patients were divided into high (blue) and low (red) gene expression groups, with patient numbers in parentheses. (a) Event-free survival (Expression cutoff: 153.312, min.grp = 8), and (b) overall survival (Expression cutoff: 137.354, min.grp = 8) curves for SFT2D1 gene expression are shown. (c) Event-free survival (Expression cutoff: 342.241, min.grp = 8), and (d) overall survival (Expression cutoff: 221.887, min.grp = 8) curves for RPS6KA2 gene expression are shown. (e) Event-free survival (Expression cutoff: 204.725, min.grp = 8), and (f) overall survival (Expression cutoff: 219.321, min.grp = 8) curves for FGFR1OP gene expression are shown.

Figure 4. Neuroblastoma high-risk patient outcomes based on T, and UNC93A gene expression using the neuroblastoma SEQC-498 patient datasets. Kaplan–Meier analyses and comparison of T, and UNC93A expression with outcome. HR-NB patients were divided into high (blue) and low (red) gene expression groups, with patient numbers in parentheses. (a) Event-free survival (Expression cutoff: 1.010, min.grp = 8), and (b) overall survival (Expression cutoff: 1.089, min.grp = 8) curves for T gene expression are shown. (c) Event-free survival (Expression cutoff: 1.086, min.grp = 8), and (d) overall survival (Expression cutoff: 1.010, min.grp = 8) curves for UNC93A gene expression are shown.

Figure 5. Neuroblastoma high-risk patient outcomes based on T, and UNC93A gene expression using the neuroblastoma TARGET-Asgharzadeh-249 patient datasets. Kaplan–Meier analyses and comparison of T, and UNC93A expression with outcome. HR-NB patients were divided into high (blue) and low (red) gene expression groups, with patient numbers in parentheses. (a) Event-free survival (Expression cutoff: 87.284, min.grp = 8), and (b) overall survival (Expression cutoff: 108.851, min.grp = 8) curves for T gene expression are shown. (c) Event-free survival (Expression cutoff: 50.677, min.grp = 8), and (d) overall survival (Expression cutoff: 50.677, min.grp = 8) curves for UNC93A gene expression are shown.

Clinical data of HR-NB patients carrying 6q loss

Clinical and demographic data for all 15 HR-NB patients carrying 6q loss are presented in Table 1. Amplification of MYCN is highly recurrent and associated with poor prognosis of NB patients, and is used to categorize the patients as high-risk, but in our dataset only 3 patients out of 15 harbored MYCN amplification. The patients of our dataset have had indeed a very poor outcome; all died of the disease.

Discussion

Outcome for children affected by HR-NB remains poor despite decades of research; then additional understanding of the pathway involved in NB pathogenesis may lead to new therapies. Chromosome signature patterns set important prognostic information for NB patients and can identify candidate tumor-related genes.²¹ Recently, it has been demonstrated that the distal 6q27 loss, containing 14 genes and spanning 7 Mb, distinguishes a group of HR-NB patients with poor survival outcomes.²⁰ In this paper we identified, using array comparative genomic hybridization, a minimal common region of loss on distal long arm of chromosome 6q27 spanning an area of 2.09 Mb. This region of loss overlaps the one found in HR-NB patients,²⁰ and harbored five genes: T, SFT2D1, RPS6KA2, FGFR1OP, and UNC93A. We found that low SFT2D1, RPS6KA2, FGFR1OP gene expression, and high T and UNC93A gene expression are associated with worse overall survival of HR-NB patients consistently across two independent NB RNA-sequencing datasets SEQC and TARGET-Asgharzadeh (http://r2.amc.nl). Low expression levels of SFT2D1, RPS6KA2, and FGFR1OP genes are associated with grim survival outcomes of HR-NB patients, suggesting that 6q27 loss may influence the prognosis through the deregulation of multiple independent pathways. These results suggested that SFT2D1, RPS6KA2, and FGFR1OP genes may act as tumor suppressor genes in human NB; however, further functional studies are required. These three genes have important functions in controlling cell growth and differentiation, and are expressed in various tissues and organs including adrenal glands. SFT2D1 gene is implicated in the fusion of transport vesicles originated from the endocytic compartment with the Golgi complex.²² RPS6KA2 gene is a putative tumor suppressor gene in epithelial ovarian cancer:²³ It is characterized by monoallelic expression in the normal ovarian epithelium, predisposing the cancer cells to complete loss of function in the context of a mutation of the single expressed allele.²⁴ FGFR1OP gene encodes a centrosomal protein that is required for anchoring microtubules to sub-cellular structures.²⁵ SFT2D1, RPS6KA2, and FGFR1OP gene expression is decreased in NB cell lines carrying 6q loss,²⁰ loss of function mutations have been described in primary HR-NB tumors,¹⁶ and low expression is associated with grim overall survival of HR-NB patients.²⁰ Moreover, FGFR1OP gene is involved in proliferation and differentiation of the erythroid lineage.²⁶ Interestingly, a recent study observed an abnormal erythrocyte maturation in children with NB.²⁷ It is possible that haploinsufficiency of FGFR1OP could be contributing to an abnormal erythropoiesis and a decreased number of circulating red blood cells as observed in NB patients.

In conclusion, we have identified a critical SRO in distal 6q27 loss that pinpointed three genes SFT2D1, RPS6KA2, and FGFR1OP that might be tightly correlated with aggressive NB. Further functional studies will be essential to confirm the presumed tumor suppressor gene(s) located within distal 6q27 region. Our findings may be translated to clinical assistance in the future, including prognosis prediction and disease prevention for HR-NB patients.

Patients and methods

Patients and tumor samples

We recovered 15 primary untreated HR-NB patients diagnosed by the Italian Association of Italian Pediatric Hematology and Oncology (AIEOP). These patients harbored tumors with chromosome 6q loss determined by array-CGH. The 15 HR-NB patients, as established by the International Neuroblastoma Risk Group (INRG) classification,²⁸ were enrolled in the HR-NBL-1 protocol (SIOPEN).²⁹ The clinical characteristics of NB patients were derived from the Italian Neuroblastoma Registry (INBR) of AIEOP, and stored in a secure system at IRCCS Giannina Gaslini, Genova, Italy. This study was approved by the Institutional Ethics Committee of Regione Liguria (Prot. n° IGG-NCA-AP-2016). Tissue samples were collected upon acceptance of written informed consent.

Genomic profile determination

We analyzed the genome-wide copy number variations for 15 frozen NB samples by array-CGH using Human Genome CGH Microarray 4x180K Kit (Agilent Technologies, Santa Clara, California, CA, USA), as previously described.³⁰

Statistical analysis

The statistical analysis was conducted with the R programming language. Differences between NB groups were showed in the figures. Experimental sample numbers (n) are pointed out in the figures, figure legends, and Results section. P values < .05 were considered statistically significant. We inquired for any association between T, SFT2D1, RP56KA2, FGFR1OP, and UNC93A gene expression and NB patient survival using two large NB tumor datasets with available survival endpoints via the R2 Genomic Analysis and Visualization Platform (http://r2.amc.nl; SEQC n = 498 and TARGET-Asgharzadeh n = 249; Figures 2–5).

Disclosure of potential conflicts of interest

The authors declare that they have no conflict of interest.

Acknowledgments

This work is supported by Cinque per mille dell’IRPEF Finanziamento della Ricerca Sanitaria, Finanziamento Ricerca Corrente Ministero della Salute, Compagnia di San Paolo, and Fondazione Italiana Neuroblastoma. The authors thank the AIEOP for tumor sample collection, and the Italian Neuroblastoma Registry (INBR). MM is a fellow of Fondazione Italiana Neuroblastoma.