The Coamplification Pattern of the MYCN Amplicon Is an Invariable Attribute of Most MYCN-Amplified Human Neuroblastomas

Amplification of the MYCN oncogene is the most extensively studied biological marker for an unfavorable prognostic outcome in patients with neuroblastoma (1–3). The overall survival probability for patients with MYCN amplification is <30% in the first 6 years after initial diagnosis (own data of 216 patients with MYCN-amplified neuroblastoma). Within this population, patients that show no metastatic disease at initial diagnosis (stages I, II, and III) respond better to current therapy strategies, including high-dose chemotherapy, autologous stem cell transplantation, and treatment with 13-cis-retinoic acid compared with patients with initial stage IV disease. Examining primary tumor samples of a large cohort of patients for coamplification pattern of different genes telomeric and centromeric to MYCN, we recently showed that the coamplification of the DDX1 gene, in proximal telomeric vicinity to MYCN, identifies a subgroup of patients with an improved survival probability compared with patients without this coamplification, independently of stage and age (4).

However, not all of the long-time-surviving patients harbored DDX1 coamplification and not every patient that underwent rapid progression of disease lacked DDX1 coamplification. Other investigators found only a trend toward a better prognosis (5) or even described a trend toward a worse prognosis for patients with DDX1-coamplified tumors (6–9). Because of the relative homogeneity regarding prognosis of the patients in these studies and the small numbers of investigated patients, statistical significance was not obtained. We hypothesized that DDX1 might influence the response of neuroblastoma cells to chemotherapy under certain conditions but does not serve as a warrant for survival per se.

In cell culture studies on methotrexate-treated Chinese hamster and mouse cells, Hahn et al. (10, 11) showed that an elevation in the amplification number of the amplified dihydrofolate reductase gene occurs under increasing selective pressure. Furthermore, the authors described a trend to a reduction in the amplicon size under these strong selective conditions. In general, the size of the amplified DNA fragments in those cell models is described to be a stable attribute independent of the manner of amplification as double-minute chromosomes or homogenously staining regions (11, 12).

These findings and our previously determined data of coamplification pattern in primary neuroblastoma prompted us to ask if the structure and the copy number of the chromosome 2p24-25 amplicon, once determined at initial neuroblastoma development, are invariable attributes for each individual MYCN-amplified neuroblastoma. To answer that question, we analyzed the coamplification pattern of subsequent tissue specimens taken at initial diagnosis and at different time intervals after treatment initiation of 33 patients with MYCN-amplified neuroblastoma using the two multiplex PCRs previously described (4). Furthermore, we determined the number of amplified copies of MYCN in relation to inhibin-β-b in a separate, semiquantitative duplex PCR (Supplementary Fig. S1).

Patients. We studied primary tumor specimens from 33 children with MYCN-amplified neuroblastomas diagnosed in Germany from 1986 to 2003. The 33 neuroblastomas were selected out of a total number of 174 tumor specimens with MYCN amplification within this period of time. The selection criterion was the availability of tumor tissue of comparable quality to different time points during therapy of each individual patient for DNA isolation.

All neuroblastoma diagnoses were confirmed by histologic assessment of a tumor specimen obtained at surgery. The tumors were classified according to the International Neuroblastoma Staging System criteria (13). All patients were treated according to previously described protocols with confirmed consent for therapy and study procedures (14, 15). Therapy included surgery, polychemotherapy, and, dependent on the randomization procedure, high-dose chemotherapy with autologous stem cell transplantation, anti-GD2 antibody, and retinoic acid treatment.

Our study group consisted of 1 stage I, 0 stage II, 10 stage III, 17 stage IV, and 5 stage IV-S tumors. The median patient age at diagnosis was 30.8 months (range 0.6-163.3 months). The median follow-up time for all 33 patients was 25.2 months (range 7.5-150.1 months). The median follow-up time of the patients that died of the disease was 18.4 months (n = 21) compared with 82.6 months for patients alive (n = 12).

The mean time interval between first and second biopsy of all investigated patients (n = 33) was 4.6 months (range 1.0-76.9 months). The mean time between first and third biopsy (n = 3) was 22.5 months. Most of the second biopsies were taken from the site of the initial tumor during operation after initial chemotherapy treatment. The mean time between first and second biopsy of these patients (n = 23) was 4.1 months (range 1.0-20.3 months). For 10 patients, recurrent disease was diagnosed based on the clinical criteria that before relapse, a full remission of the initial disease occurred during therapy. The mean time between first and second biopsy of these patients (n = 10) was 17.2 months (range 5.2-76.9 months). Clinical data and amplification pattern of all investigated patients are presented in Table 1.

Methods. Isolation of genomic DNA and the two multiplex PCR was done as previously described (4). The mean tumor cell content of investigated MYCN-amplified neuroblastoma in the last 6 years in our laboratory was 80% (range 98% maximum to 60% minimum, n = 173). The mean tumor cell content of the investigated specimens in this study was 80% (range 95% maximum to 65% minimum) at initial diagnosis. The mean tumor cell content of the investigated specimens was 80% (range 90% maximum to 60% minimum) at second biopsy (Table 1).

For quantification of the MYCN copy number per haploid genome (Mcn/hg), a separate duplex PCR approach was established. Because the band intensities of the investigated genes were, in addition to the amplicon number, dependent on the coamplification pattern in both multiplex PCR, it was essential to design a separate duplex PCR for determination of the MYCN copy number relative to a control gene (inhibin-β-b) located on the same chromosome (primer used: MYCN-upper, CATCCACCAGCAGCACAACTATG; MYCN-lower, CCAGAGGCTCCCAACCGT-CAC; inhibin-β-b-upper, CATTGCCTTGTGTTCTCCTT; inhibin-β-b-lower, GAATGCGGTGCCTG-CTGTC; denaturation 20 s, 95°C; annealing 20 s, 63°C; elongation 60 s, 72°C). The amplified PCR fragments (10 μL of the PCR reaction) were separated on a 2% agarose gel by electrophoresis, visualized by ethidium bromide staining and UV light exposure, and finally documented and analyzed with the Image Master VDS software [GE-Healthcare(Amersham-Pharmacia), Munich, Germany]. The relative level of the PCR bands was determined by densitometry. Moreover, a ratio of the MYCN band intensity to the intensity of the inhibin-β-b band was determined.

A standard curve was created using a dilution series of genomic DNA from a neuroblastoma with high copy number MYCN-amplified neuroblastomas, with known diploid karyotype and known Mcn/hg determined by Southern blot as described previously (4). The tumor DNA (∼250 Mcn/hg) was diluted stepwise with human placental DNA; thus, the Mcn/hg was halved with each dilution step. The obtained standard curve was used to deduce the Mcn/hg for each patient's tumor specimens (Supplementary Fig. S1).

The coamplification pattern of the investigated genes on chromosome 2p24-25 and the Mcn/hg for each investigated tumor are given in Table 1.

Statistics. Fisher's exact test was used to compare the amplification status for each gene determined at initial diagnosis and at later time points. The paired t test was used to compare the Mcn/hg determined at initial diagnosis and at later time points. Statistical analyses were done with the SPSS version 10.0 software (SPSS, Inc., Chicago, IL). P values of ≤0.05 were regarded as significant.

The coamplification pattern within the MYCN amplicon is an invariable attribute of most neuroblastoma tumors. In the present cohort besides MYCN amplification, 15 (46%) of the investigated tumors showed coamplification for DDX1, 11 (33%) for NAG, 5 (15%) for NSE1, 2 (6%) for LPIN1, 31 (94%) for EST-AA581763, 2 (6%) for SMC6, and 1 (3%) for SDC1 (Fig. 1; Table 1) The coamplification of genes located telomeric to MYCN (DDX1, NAG, and NSE1) was less frequent compared with our previously published data that were determined on a larger cohort (n = 98 patients; ref. 4). However, the differences were not statistically significant and could be explained by the smaller patient numbers in the present study and the relatively high number of patients with recurrent disease selected for investigation within this study.

In the cohort of 33 patients investigated in this study, all patients except one did not show any change in the coamplification pattern during the course of disease, which means that the amplicon structure was invariable at the different time points of subsequent biopsies.

Patient 18 showed a loss of primarily coamplified genes (NSE1, NAG, and DDX1) within the amplicon at the second biopsy date, 76.9 months after the initial diagnosis. This patient was initially diagnosed as a stage I tumor at the age of 6.6 months. At the age of 6.4 years (83.5 months), a recurrent tumor was diagnosed that again was defined as stage I. To the date of our investigation, the patient was still alive (114 months after initial diagnosis; Fig. 2).

In three patients, three consecutive tumor biopsies were investigated. In DNA isolated from third biopsy tissue (n = 3), no change in the amplicon structure was found compared with the first or second biopsy in all three patients. Coamplification of DDX1 was found in only one of these three specimens. No tumor showed amplification of genes located further telomeric to MYCN (Table 1).

The number of amplified MYCN copies decreases after initial chemotherapy and increases in recurrent disease. Unlike the coamplification pattern, a significant change in amplification number was observed in the investigated cohort (Fig. 3).

Thus, we found the Mcn/hg to be decreased after initial chemotherapy (n = 23; Fig. 3A). The median Mcn/hg at initial diagnosis was 25 Mcn/hg compared with 20 Mcn/hg after initial chemotherapy (25-75% percentile: 25-40 Mcn/hg compared with 20-30 Mcn/hg, respectively).

In contrast, we found an increase of Mcn/hg in tumor samples of recurrent disease compared with their primary counterparts (n = 10; Fig. 3B). The median MYCN copy number at primary diagnosis was 27 Mcn/hg compared with 46 Mcn/hg in recurrent disease (25-75% percentile: 10-36 Mcn/hg compared with 30-64 Mcn/hg, respectively). Both alterations are statistically significant (P < 0.05, paired t test).

Not much is known about the initiation and genesis of the MYCN amplicon in neuroblastoma development. Cheng et al. (16) revealed that, preferentially, the paternal allele undergoes the amplification process. Referring to patient age, recent data indicate that amplification of MYCN occurs to a defined point of time in neuroblastoma development as 65% of patients with MYCN-amplified neuroblastomas are between 12 and 40 months of age (17). Our data from about 247 patients are in consent with this age distribution, as, in our cohort, 55% MYCN-amplified tumors are initially diagnosed at an age between 12 and 40 months.

The amplicon size varies between 350 and >1,000 kb in length, including different coamplified genes telomeric and centromeric of MYCN (4, 18). The amplicon number varies from 2 to 256, and, in some cases, more copies per haploid genome in neuroblastoma cells at that time of diagnosis. In the great majority of MYCN-amplified neuroblastomas, the amplified genomic regions are initially present in the form of extrachromosomal double minutes rather than in homogeneously staining regions, which can be found mainly in recurrent disease (19, 20).

Extensive DNA rearrangements can be found within the amplified region, leading in part to the phenomenon that coamplified genes might be presented not equally in copy number compared with MYCN in either homogenously staining regions or double-minute chromosomes (21, 22).

Yoshimoto et al. (20) described the possibility of coexisting double-minute chromosomes and homogenously staining regions in single tumor cells besides cells with either double-minute chromosomes or homogenously staining regions in one neuroblastoma tumor. They claim a possible transition from double-minute chromosomes to homogenously staining regions during tumor development based on the finding that cells that reintegrate amplified oncogenes into the genome show a growth advantage over cells harboring the amplified gene copies in extrachromosomal double-minute chromosomes (23).

A change or evolution of the amplicon structure has not yet been reported for neuroblastoma. Our findings show that a change of the amplicon structure is uncommon during the course of disease but is possible in neuroblastoma recurrent disease. In the great majority of cases, we found the amplicon structure to be an invariable attribute of the individual tumor that does not undergo rearrangement during therapy. The observed change in the coamplification pattern in one patient could be explained by subsequent selection of more therapy-resistant neuroblastoma cell clones harboring the MYCN amplicon of smaller extent (Fig. 2B). As a prerequisite for this explanation, a heterogeneous ancestor cell population, which contains cells harboring MYCN amplicons of different sizes at determination of the individual neuroblastoma, has to be presumed. Alternatively, a successive change of the amplicon structure, caused coincidentally in highly proliferating cells or by DNA interfering agents and favoring cells with amplicons of smaller size over a given period of time, could be a possible mechanism. The latter explanation is more consistent with observations in common cell culture models (reviewed in ref. 23).

We previously showed coamplification and overexpression of DDX1 to occur preferentially in long-time-surviving patients and discussed a possible influence on the response to chemotherapeutics (4). Hypothetically, cells that lose this coamplification telomeric to MYCN should become more resistant to chemotherapeutic treatment. Our finding in the neuroblastoma of patient 18 that lost coamplification of NSE1, NAG, and DDX1 in the relapsing tumor is consistent with this hypothesis.

Interestingly, for amplified copies of the c-myc and MYCN genes occurring in double-minute chromosomes, a reduction in copy number after treatment with hydroxyurea can be observed, but not for copies in homogenously staining regions. After hydroxyurea treatment, the amplicon number can recur to the initial count (24). These findings show a possibility to influence gene amplification by medical treatment but also indicate that these exogenously induced genomic changes are not irreversible. In 1987, Brodeur et al. (25) investigated consecutive tumor samples of 12 patients with MYCN-amplified neuroblastomas and describe no change in the amplicon number for their investigated cohort. However, considering the given data of each individual patient, the findings of Brodeur et al. are strongly consistent with our findings. In their study, the Mcn/hg was found to be increased in tumor tissue of four of the seven patients with recurrent disease when compared with the primary tumor tissue. In consecutive specimens of patients during treatment, the Mcn/hg decreased in three of four reported patients.

These findings and our own observation of a reduction of the Mcn/hg after initial chemotherapeutic treatment indicate that the reported observations in cell culture regarding elimination of extrachromosomal amplified MYCN in response to treatment is also a possible mechanism of adaptation in primary MYCN-amplified neuroblastomas. In a cellular context, this regulation is reasonable as cells that eliminate extrachromosomal copies of MYCN might decelerate MYCN-induced proliferation. This is an observed mechanism to avoid massive apoptosis triggered by chemotherapy and enforced by myc oncogenes (26). As we lack the information if MYCN amplification is present in double-minute chromosomes or homogenously staining regions in neuroblastomas of our investigated patients, a correlation of the molecular phenotype of amplification to a decline of the amplification number is not possible. On the other hand, we observed a significant increase in the Mcn/hg in recurrent disease compared with the initially diagnosed primary neuroblastoma. These cells might represent highly proliferating clones that could bypass myc-enforced apoptosis and, thus, have benefit from harboring a higher Mcn/hg. We have to annotate that the observed changes of the Mcn/hg could also result from a different content of viable tumor cells within the subsequent tissue specimens, based on the used PCR method. However, the tumor cell content of the investigated specimens in this study is interindividually and intraindividually comparable (Table 1).

The observed changes in the amplicon copy number over time might reflect abided regulatory mechanisms in neuroblastoma cells comparable with cell culture models of acquired gene amplification. In contrast to an acquired gene amplification under selective pressure, a reason for an initialization of the MYCN amplification in developing neuroblastic cells, with the possibility to lead to MYCN-amplified neuroblastoma, remains unclear. Once initiated, the MYCN amplification is found to be irreversible in neuroblastoma tumors, and, once determined at time of initial amplification, the structure of the MYCN amplicon is invariable during the course of disease, including chemotherapy with DNA-damaging agents, in the majority of neuroblastoma tumors.

A fundamental question arises from our findings. Is there a cellular program superior to MYCN amplification that advises a cell, deriving from the neural crest, to amplify MYCN in a certain developmental condition, which then could lead to MYCN-amplified neuroblastoma if this genomic condition becomes an irreversible attribute?

We thank Monika Faulhaber for performing many of the PCR experiments, our colleagues from about 100 German pediatric centers for providing us with neuroblastoma tumor material since 1981, Prof. Berthold and Dr. Hero (University of Cologne, Cologne, Germany) for providing clinical data, and Prof. Lampert (University of Gießen, Gießen, Germany) for support of our research group.