Neuroblastoma with flat genomic profile: a question of representativity?

Abstract

Neuroblastoma is one of the most common paediatric malignancies. Detection of somatic genetic alterations in this tumour is instrumental for its risk stratification and treatment. On the other hand, an absence of detected chromosomal imbalances in neuroblastoma biopsies is difficult to interpret because it is unclear whether this situation truly reflects the tumour genome or if it is due to suboptimal sampling. We here present a neuroblastoma in the left adrenal of a newborn. The tumour was subjected to single-nucleotide polymorphism array analysis of five tumour regions with >80% tumour cells in histological mirror sections. This revealed no aberrations compared with a normal reference sample from the patient. Whole exome sequencing identified two single-nucleotide variants present in most tumour regions, corroborating that the tumour resulted from monoclonal expansion. Our data provide proof-of-principle that rare cases of neuroblastoma can have a normal whole genome copy number and allelic profile.

Background

Neuroblastoma (NB) is the most common malignant paediatric solid tumour outside the central nervous system and the most common malignancy in newborns.¹ NB cells simulate the morphology of fetal neuroblasts and the tumour is assumed to arise from neural crest progenitors.² Accordingly, its most common sites of origin include the adrenal medulla and the sympathetic ganglia. The disease course of NB exhibits a staggering versatility, which ranges from spontaneous maturation to progressive, lethal disease despite aggressive multimodal therapy. Ever since amplification of the proto-oncogene MYCN was shown to be a marker for high-risk cases,^{3 4} genetic analysis of tumour cells has been crucial for correct risk stratification and treatment of NB. Besides overall ploidy and chromosomal rearrangements such as deletions of 1 p and 11q, which signify distinct patterns of treatment response, there are also mutations in single genes that signal response to particular drug protocols.⁵ Examples are anaplastic lymphoma kinase inhibitors for NBs that exhibit mutation/amplification of the corresponding gene and mitogen-activated protein kinase inhibitors for tumours with somatic mutations in the RAS-MAPK oncogenic signalling pathway.

Few if any clinical correlates have been reported for NBs that fail to show genetic aberrations even at high-resolution pan-genomic screening—today typically performed up-front by single-nucleotide polymorphism (SNP) array or in a retrospective setting by next-generation whole exome and/or whole genome sequencing. Hence, it remains unknown how common such cases are. It is also unexplored whether NBs with a normal genetic allele/copy number profile constitute a specific clinical–genetic subgroup, as has been shown for acute myeloid leukaemias where normal karyotypes are associated with the presence of specific mutations at the nucleotide level.⁶ This raises the question whether NB samples where genomic aberrations are not found (flat-line genomic profiles) can ever be considered representative of the neoplastic cell population or if they invariably correspond to mis-sampling of, for example, tumour capsule, areas dominated by inflammatory cells or tumour haemorrhage. The latter assumption could lead to unnecessarily ambitious sampling attempts in order to find genetic markers that are simply not there. The question of representativity in cases of NB with flat genome profiles is therefore not trivial from a clinical point of view. We here describe a proof-of-principle NB case where meticulous multiregional tumour sampling followed by high-resolution SNP array failed to detect allelic/copy number imbalances.

Case presentation

The patient was the offspring of two healthy non-consanguineous parents. He was born with APGAR score 8-10-10 at gestational week 42+1 through vaginal delivery after an uneventful pregnancy. At neonatal routine health control the day after birth, a non-translucent, dark scrotum was observed, raising the suspicion of scrotal haematoma and testicular torsion. Surgical exploration confirmed fresh intrascrotal haemorrhage but showed no torsion. Abdominal ultrasound revealed a 6.5×5×4.5 cm mass in the left adrenal, consisting of fresh haemorrhage with a juxtaposed adrenal tumour. The suspicion of NB was supported by strong uptake in the region of the left adrenal on iodine-123-meta-iodobenzylguanidine (MIBG) scan. At 6 weeks of age, the patient underwent an open left-sided adrenalectomy with en bloc resection of the haematoma with surrounding tumour.

Investigations

The cut surface of the adrenalectomy specimen revealed a central, encapsulated haematoma measuring 3 cm in largest diameter (figure 1A). This was surrounded by a solid, congested tumour mass, 5 cm in largest diameter. There was no macroscopically visible tumour necrosis. The tumour capsule was intact and adhered to small patches of periadrenal fat. At light microscopy, the tumour consisted of islands of small round blue cells with neuroblastic features, having scant cytoplasm and salt and pepper chromatin (figure 1B), partially surrounded by fresh haemorrhage (figure 1C:I). The neuroblastic cells infiltrated the adrenal cortex (figure 1C:II) and exhibited perivascular growth around subcortical vessels (figure 1C:III). The tumour cells were positive for chromogranin at immunohistochemistry (figure 1C:IV) and grew in a highly vascularised neurofibrillary matrix (figure 1C:V, VI). There was no microscopic invasion of the tumour capsule or surrounding tissues. No Schwannian stroma or ganglion cell differentiation was observed. The mitosis–karyorrhexis index was low and the tumour was classified as a low-risk NB according to the international NB pathology classification system.⁷

Figure 1.

A NB with flat genomic profile. (A) Red-brown cut surface of an adrenal Tu surrounding a core of haemorrhage (h). (B) The tumour was divided into sectors a–e, each of which was bivalved for histological preparation (surrounding panels) and genetic analyses, respectively. (C) The tumour consisted of islands of small round blue cells with nb features (I), partially surrounded by fresh haemorrhage beneath the ac. NB cells focally infiltrated the adrenal cortex (II) with perivascular growth around subcortical ar (III) and ve (III). NB cells were positive for chromogranin (IV), grew in nests surrounded by thin capillaries (V), and exhibited a neurofibrillary background matrix (VI). (D) Whole genome copy number profile of tumour samples a–e and a N control sample from surrounding fatty tissue; for each sample, a series of data points reflect logarithm 2 of copy number values of probes located from Chr 1 to the Y chromosome. (E) DNA index by flow cytometry of tumour sample a shows a diploid profile (2 n) with a minimal tetraploid (4 n) G2/M peak. (F) Allele profile from sample a, showing a balanced status including heterozygous (A–B=0) and homozygous (2A/2B=1/–1) alleles. Note that sex chromosomes are not balanced as the patient was male (XY). ac, adrenal cortex; ar, arterioles; Chr, chromosome; N, normal; nb, neuroblastic; NB, neuroblastoma; Tu, tumour; ve, venules.

For SNP array analysis, a central section covering the largest transversal surface was cut from the fresh tumour specimen, to be further subdivided into five ~1×1×1 cm³ samples. Each of these were then cut in two halves, of which one was formalin-fixed and paraffin-embedded for histological studies, while the other was subjected to DNA extraction according to standard methods (DNeasy Blood and Tissue Kit, Qiagen). By light microscopy, >80% of nucleated cells in each of the formalin-fixed samples were tumour cells. Extracted DNA from all five fresh tumour samples was subjected to whole genome genotyping by the Cytoscan HD SNP array (Affymetrix/ThermoFisher) as described.⁸ This did not reveal any allelic imbalances at employment of standard cut-off values (>10 consecutive markers with log2 >0.05 or <−0.10; d homozygous blocks >5 Mb), or at comparison with SNP array data from DNA extracted from the patient’s periadrenal fat (figure 1D and F). DNA index determined by flow cytometry of one tumour sample⁹ showed a pure diploid DNA content (figure 1E).

For whole exome sequencing, freshly cut 10 µm sections from the five formalin-fixed tumour samples were subjected to paraffin removal and DNA extraction, followed by library preparation by the SureSelectXT Human All Exon Kit (V.6; Agilent; paired end with read length 2×125 bp), followed by sequencing by Illumina technology (GATC Biotech). Mean coverage was in the range of 129–160× in the tumour samples. Analysis of the sequencing data was performed as described.¹⁰ After filtering against normal DNA from the patient and manual inspection of putative variant calls, two nucleotide variants remained that were ascertained independently by having ≥10 alternate reads in ≥2 samples. These were both synonymous single-nucleotide variants. XPC c.G267A was detected in all five tumour samples at a variant allele frequency (VAF) of 0.26–0.45, while SPTBN4 c.G276A was found in four samples at a VAF of 0.08–0.26.

Outcome and follow-up

The patient did not receive any additional specific treatment after surgery but was followed at regular intervals. Two years after surgery, he is tumour-free, without complications from surgery and follows developmental milestones as expected.

Discussion

NBs make up approximately 50% of congenital malignant solid tumours.^{1 11} In newborns, they are typically confined to the adrenal gland with approximately half of the cases showing localised (stage I-II) disease akin to the present case. Furthermore, more than half of patients with congenital NB with disseminated disease have the special 4S (MS) type, prone to spontaneous regression. Survival of congenital NB is excellent. In a recent retrospective study, the overall survival for neonatal NB was >90% with the adverse prognostic factor of MYCN amplification ascertained in only 2% of the cases.¹¹ The few congenital NBs included in recent sequencing studies have indicated that also other adverse genetic markers such as 1 p deletion and chromothripsis are rare.¹² However, to our knowledge, even for this patient group there are few if any reports of NBs that fail to exhibit large-scale genomic alterations. One obvious reason for this is the difficulty of excluding that the lack of genetic anomalies is simply the result of erroneous sampling, leading to analysis mostly of the patient’s non-neoplastic cells. For state-of-the-art risk stratification and individualised treatment, molecular analyses of NB should include at least evaluation of the absence or presence of MYCN amplification. This is routinely done by fluorescence in situ hybridisation or SNP array, and less commonly by other methods such as next-generation DNA sequencing or multiplex ligation-dependent probe amplification. To achieve rapid and reliable results, these analyses are typically performed on fresh or freshly frozen tumour tissue. However, only in rare cases, such as the one presented here, is there a careful morphological quality control ensuring that the analysed tissue actually represent viable tumour. Without such correlation between histopathology and genetic data, it is impossible to correctly interpret the finding of a normal copy number/allelic profile.

The present tumour was subjected to ambitious multiregional sampling with microscopic analysis of the areas immediately next to those subjected to genetic analysis, in turn making it possible to rule out the alternative that the analysed DNA was not representative of tumour. This leaves the alternatives that the lesion in question was hyperplastic rather than neoplastic in nature. The former would be consistent with its flat whole-genomic profile, but it fits poorly with its invasive histology and also with the finding of somatic mutations by whole exome sequencing. Of these, XPC c.G267A exhibited VAFs that would fit well with a prevalence in all tumour cells in all analysed samples. In contrast, the lower VAF values for SPTBN4 c.G276A together with its absence in one of the tumour samples is indicative of a subclonal origin, that is, representing a more recent mutation that still occurred sufficiently early to disseminate over several sampled regions. However, both variants were synonymous and are therefore most likely passengers rather than drivers of tumourigenesis. The genetic analyses thus failed to give any clues to the pathogenesis of the present case. This leaves its mechanistic basis open to speculation, including microenvironmental factors, epigenetic mechanisms, or mutations other than point mutations or allelic imbalances. Examples of the latter are balanced translocations leading to gene fusions or promoter swapping.

In summary, this patient demonstrates that a normal whole genome profile can truly represent NB tumour cells; it does not by necessity mean that sampling has failed. The latter insight is important to avoid overzealous sampling with the associated risk of delayed treatment, unnecessary pain and other complications for the patient.

Learning points.

• Even though neuroblastomas presenting in newborns have a favourable outcome, genetic analysis of tumour cells is important to find rare high-risk patients and treat them accordingly.

• The finding of a normal allelic and copy number profile at high-resolution genomic analysis does not by necessity result from misrepresentative sampling; in some cases, it can be a true representation of neuroblastoma cells.

• Whether neuroblastomas with a normal genomic profile correlate to any specific clinical characteristics remains to be established; such studies mandate careful pathology workup to ascertain sample representativity.