Age dependence of tumor genetics in unfavorable neuroblastoma: arrayCGH profiles of 34 consecutive cases, using a Swedish 25-year neuroblastoma cohort for validation

Neuroblastoma is a childhood malignancy that arises from embryonic cells of the sympathetic ganglia or the adrenal medulla [1]. It is mainly a disease of infants and toddlers; more than half of patients with neuroblastoma are diagnosed before two years of age and ~90 percent before age six [2, 3]. This age dependence of the incidence of neuroblastoma may be a consequence of developmentally determined disappearance of the pool of immature cells from which neuroblastomas are thought to derive. The disease is clinically diverse, and ranges from cases with a very dismal prognosis, despite modern intensive multimodal therapy, to those with an excellent chance of survival [2]. This variation in clinical behavior is also highly age dependent: children who are diagnosed after two years of age suffer predominantly from aggressive forms. Neuroblastomas that are diagnosed during adolescence and young adulthood are rare, but they are of particular concern because they almost invariably progress, although with an indolent course [4, 5]. In sharp contrast, tumors that are diagnosed before 18 months of age are generally associated with a favorable prognosis. Such tumors are usually less advanced, have a propensity for spontaneous involution or maturation, and respond well to mild chemotherapy [2].

These clinical differences correspond to clear differences in tumor genetics [2]. As a general rule, prognostically favorable tumors display numerical imbalances of entire chromosomes and have near-triploid DNA content, whereas higher-risk tumors present with segmental chromosomal aberrations (SCAs) and are often pseudo-diploid. MYCN gene amplification (MNA) was one of the first genetic markers for highly aggressive neuroblastoma to be established [6], and remains a powerful prognostic indicator [7]. More recently, an independent prognostic value of a segmental deletion of 11q has also been recognized [7‐9]. Both these aberrations are incorporated in the present International Neuroblastoma Risk Group (INRG) classification system for treatment stratification [10]. Several arrayCGH studies in the recent years support the MNA/segmental 11q loss dichotomy of high-risk neuroblastoma and indicate that any type of segmental numerical chromosomal aberration is a negative prognostic sign [11‐13]. However, the representativeness of the studied tumor materials may be questioned because the tumors were from multiple sources and, hence, selection bias may have occurred. Therefore, the present study aims at characterizing the heterogeneity of genetic aberrations in aggressive neuroblastoma by exploiting a consecutive, population-based series of tumors, the representativeness of which was tested against data in the Swedish Childhood Cancer Registry. The most striking observations were related to age at tumor presentation: MNA tumors were associated with a particularly early age at diagnosis and low numbers of other chromosomal aberrations suggesting a rapid tumor evolution with few genetic hits involved, whereas 11q deleted tumors were diagnosed at older ages and showed significantly more SCAs, the numbers of which were positively correlated with the age at diagnosis, suggesting a chromosomal instability phenotype with a more stepwise tumor evolution. Other tumors seemed to be the result of late progression of low-risk neuroblastoma or of gene amplifications other than MYCN. This clinicogenetic diversity of unfavorable neuroblastoma is likely to reflect differences in tumor evolution and growth, which may have therapeutic implications.

In this report, we describe the DNA copy number profiles of a consecutive series of neuroblastomas that were selected on the basis of unfavorable characteristics. The findings revealed considerable genetic heterogeneity within this clinically troublesome group, which was particularly evident when comparing tumors with MNA to those with segmental 11q deletions. With few exceptions, MNA and segmental 11q loss were mutually exclusive and defined two genetic subgroups of equal size that comprised more than three-quarters of the total samples. Such genetic dichotomy of advanced neuroblastoma has been well described previously [2, 7, 8, 11, 15] and both MNA and segmental 11q loss are included in the current INRG algorithm for pretreatment stratification of risk [10]. Less predictably, we also observed a clear clinical difference between these two genetic subgroups in relation to age: MNA tumors affected the youngest children of the series. It is surprising that this age dependence with respect to the tumor genetics of neuroblastoma has not received much scientific attention previously, although mentioned in several previous studies [9, 11, 15, 25]. In view of the relatively moderate size of the present tumor series, it was important that we were able to confirm a generally low age at diagnosis for children with MNA tumors using independent data from the Swedish Childhood Cancer Registry; these data argued clearly against a bias in the present material. We conclude from the present findings that unfavorable neuroblastomas are predominantly of the MNA type when diagnosed under the age of 2 years, whereas tumors with loss of 11q and other genetic variants predominate after 3.5 years of age.

As the Swedish Childhood Cancer Registry, due to lack of records, could not be used to verify the older age at diagnosis for children with 11q-deleted tumors we searched the literature for this information: Spitz et al. [9] reported on segmental 11q deletions from a cohort of 611 neuroblastomas, found in 159 tumors. The median age at diagnosis of these 11q-deleted tumors was 3–5 years, constituting 59 percent of the tumors of this age range. Michels et al. [11] reported 48 and 28 months as median ages at diagnosis for ten 11q-deleted and 22 MNA tumors, respectively. In a meta study by Vandesompele et al. [25] poor risk neuroblastomas were separated into two genetic “clusters”: The median age of 45 children belonging to the 11q deletion “cluster” was 41 months, although notably only 30 of the tumors were actually 11q deleted. The corresponding figures for 74 children belonging to the MNA “cluster” was 26.9 months median age with actual MNA seen in 51 tumors. Finally, Carén et al. [15] reported a median age of 42 months for 21 children with 11q-deleted tumors (four of which were common with our study) compared to 21 months for 37 children with MNA tumors (seven in common with our study). Together, these data consistently support an age difference between the two groups, although a certain bias towards older ages for 11q-deleted tumors in the present study is observed (58.5 months median age).

Another difference that was observed between MNA and 11q-deleted tumors was the higher number of SCAs in the later group of tumors and a dependence of their prevalence on age, both of which imply a chromosomal instability phenotype, as suggested previously by Carén et al. [15]. Future analyses using larger sample sizes, including multiple tumor specimens (synchronous and metachronous), might be helpful in confirming such chromosomal instability. Also, deep sequencing might shed additional light on differences in genomic integrity at the DNA level among subsets of neuroblastoma. A recent report [26], based on whole-genome sequencing of 87 neuroblastomas of all stages, showed that the frequency of somatic amino-acid-changing mutations strongly correlated to tumor stage, survival, and age at diagnosis. However, no difference in frequency of such mutations was detected when comparing MNA tumors to high-stage non-MNA tumors. Hence, this aformentioned study reported a general age dependence of amino-acid-changing mutations in neuroblastoma, albeit the specific issue of DNA instability in tumors with deleted 11q, as compared to MNA tumors, was not addressed.

On the basis of the profile subtypes that were observed in the presently relatively limited series of tumors, and in view of previous data by others, we speculate on the existence of four genetic routes for the genesis of aggressive neuroblastoma:

On the basis of these categories, we suggest the following metaphor for the genesis of unfavorable neuroblastoma: MNA, segmental deletion of 11q, and low-risk precursor lesions provide an elevator, a staircase, and a first step of a staircase, respectively, towards unfavorable neuroblastoma. This metaphor builds on the notion that the first steps in tumor development take place during a common early developmental phase when the immature cells that are the source of neuroblastoma are still present in the sympathetic nervous system, whereas subsequent tumorigenic events differ in malignant effect between the genetic routes. Our speculations on main routes for the pathogenesis of unfavourable neuroblastoma are summarized in Figure 6, showing the four mentioned routes and adding also adolescent neuroblastoma as a separate, but still obscure, route. The present delineation of genetic subsets of unfavorable neuroblastoma might have therapeutic implications. It would seem that MNA tumors are more proliferative, which calls for a dose-intensive treatment, and the relatively “clean” genetics of MNA neuroblastomas would appear to make them suitable for tailored treatments that target the functions of MYCN, e.g. apoptosis [35]. Novel therapeutic strategies specifically for 11q-deleted neuroblastoma are urgently needed, and the identification of novel subsets of neuroblastomas with non-MYCN amplifications may also have therapeutic consequences, but these will have to await more conclusive information on the genetic and clinical homogeneity of such tumors. Finally, the possibility of a transition from low- to high-risk disease in some cases may have implications for the monitoring of low-risk rest tumors.

CC held a postdoc research position at the Department of Immunology, Genetics and Pathology, Uppsala University during the work. TM is a Professor of tumor genetics at the Department of Clinical Genetics, University of Gothenburg and a senior researcher on neuroblastoma. JS was a PhD student at the Department of Immunology, Genetics and Pathology, Uppsala University involved in the development of the BAC array platform and later holds a postdoc position at the Department of Oncology-Pathology, Karolinska Institutet. CT is an MD, PhD, specialist in pediatric oncology and a senior researcher on neuroblastoma at Karolinska Institutet. PK is an MD, PhD, and specialist in pediatric oncology with national responsibilities for neuroblastoma treatment, and senior researcher on neuroblastoma at Karolinska Institutet. JD is a professor of experimental pathology at the Department of Immunology, Genetics and Pathology, Uppsala University. TDS is an associate Professor at the Department of Oncology-Pathology, Karolinska Institutet and formerly at the Department of Immunology, Genetics and Pathology, Uppsala University. FH is an MD, PhD, and specialist in pediatric oncology and senior researcher on neuroblastoma at the Department of Immunology, Genetics and Pathology and Women’s and Children’s Health, Uppsala University.