Keywords

Introduction

During normal development and self-renewal, cells evolve to perform highly specialized functions to meet the physiologic needs of the organism. These processes involve tightly regulated activities that include continued cell proliferation, differentiation into specialized cell types, and programmed cell death (apoptosis). An intricate system of checks and balances ensures proper control over these physiologic processes. The genetic composition (genotype) of a cell forms the basis for that control, but the environment also plays a crucial role in influencing cell fate. Cells use complex signal transduction pathways to sense and respond to neighboring cells and their extracellular milieu. In addition, environmental factors may have a direct impact on cell phenotype and fate by causing DNA damage that permanently alters the host genome.

Cancer is a genetic disease whose progression is driven by a series of accumulating genetic changes influenced by hereditary factors and the somatic environment. These genetic changes result in individual cells acquiring a phenotype that provides those cells with a survival advantage over surrounding normal cells. Our understanding of the processes that occur in malignant cell transformation is increasing, with many discoveries in cancer cell biology having been made using childhood tumors as models.

Cell Fate

Stem Cells

The development and maintenance of the tissues that comprise an organism are driven by stem cells. These are cells with the potential for both self-renewal and terminal differentiation into one or more cell types. They, therefore, play a critical role in normal tissue turnover and repair. The fate of most of these stem cells is generally one of terminal differentiation and either quiescence or apoptosis. However, a small percentage of stem cells maintain their pleuripotent capacity. It is becoming increasingly recognized that these same stem cells that are essential for maintaining an organism are also central to the development of malignancy and therapy resistance [133]. Cancer stem cells, like normal stem cells, possess remarkable proliferative and self-renewal capacities, while the larger portion of partially differentiated tumor cells possess quite limited reproductive potential.

Programmed Cell Death

Multicellular organisms have developed a highly organized and carefully regulated mechanism of cell death in order to maintain cellular homeostasis. Normal development and morphogenesis are often associated with the production of excess cells, which are removed by the genetically programmed process called apoptosis. Apoptosis is a highly regulated event which can be effected by either death receptor-mediated or mitochondrial pathways by activating specific signaling molecules. Both pathways converge onto a group of effector caspases, leading to morphologic and biochemical changes characteristic of apoptosis. Cells undergoing apoptosis have distinct morphologic features (plasma membrane blebbing, reduced volume, nuclear condensation), and their DNA is subjected to endonucleolytic cleavage.

Receptor-mediated apoptosis is initiated by the interaction of “death ligands” such as tumor necrosis factor α (TNFα), Fas, and TNF-related apoptosis-inducing ligand (TRAIL) with their respective receptors. This interaction is followed by aggregation of the receptors and recruitment of adaptor proteins to the plasma membrane, which activate caspases [92]. Caspases are a large family of proteases that function in both the initiation of apoptosis in response to proapoptotic signals and in the subsequent effector pathway that disassembles the cell. Thus, apoptosis limits cellular expansion and counters cell proliferation. Because cell survival signals may also be activated through parallel pathways, the fate of a cell is determined by the balance between death signals and survival signals [65]. Other signals arising from cellular stress (e.g., DNA damage, hypoxia, oncogene activation) may also effect cell cycle arrest or apoptosis.

An alternative to cell death mediated by receptor-ligand binding is cellular senescence, which is initiated when chromosomes reach a critical shortened length. Eukaryotic chromosomes have DNA strands of unequal length, and their ends—telomeres—are characterized by species-specific nucleotide repeat sequences. Telomeres stabilize the ends of chromosomes, which are otherwise sites of significant instability [121]. Over time and with each successive cycle of replication, chromosomes are shortened by failure to complete replication of their telomeres. Thus, telomere shortening acts as a biologic clock, limiting the lifespan of a cell. Germ cells, however, avoid telomere shortening by using telomerase, an enzyme capable of adding telomeric sequences to the ends of chromosomes. This enzyme is normally inactivated early in the growth and development of an organism. Persistent activation or the reactivation of telomerase in somatic cells appears to contribute to the immortality of transformed cells.

Malignant Transformation

Alteration or inactivation of any of the components of normal cell regulatory pathways may lead to the dysregulated growth that characterizes neoplastic cells. Malignant transformation may be characterized by cellular dedifferentiation or failure to differentiate, cellular invasiveness and metastatic capacity, and/or decreased drug sensitivity. Tumorigenesis reflects the accumulation of excess cells that results from increased cell proliferation and decreased apoptosis or senescence. Cancer cells do not replicate more rapidly than normal cells, but they show diminished responsiveness to regulatory signals. Positive growth signals are generated by proto-oncogenes, so named because their dysregulated expression or activity can promote malignant transformation. These proto-oncogenes may encode growth factors or their receptors, intracellular signaling molecules, and nuclear transcription factors (Table 3.1). Conversely, tumor suppressor genes, as their name implies, control or restrict cell growth and proliferation. Their inactivation, through various mechanisms, permits the dysregulated growth of cancer cells. Also important are the genes that regulate cell death. Their inactivation leads to resistance to apoptosis and allows accumulation of additional genetic aberrations.

Table 3.1 Proto-oncogenes and tumor suppressor genes in pediatric malignancies
Full size table

Cancer cells carry DNA that has point mutations, viral insertions, or chromosomal or gene amplifications, deletions, or rearrangements. Each of these aberrations can alter the context and process of normal cellular growth and differentiation. Although genomic instability is an inherent property of the evolutionary process and normal development, it is through genomic instability that the malignant transformation of a cell may arise. This inherent instability may be altered by inheritance or exposure to destabilizing factors in the environment. Point mutations may terminate protein translation, alter protein function, or change the regulatory target sequences that control gene expression. Chromosomal alterations create new genetic contexts within the genome and lead to the formation of novel proteins or to the dysregulation of genes displaced by aberrant events.

Genetic abnormalities associated with cancer may be detected in every cell in the body or only in the tumor cells. Constitutional or germline abnormalities are either inherited or occur de novo in the germ cells (sperm or oocyte). Interestingly, despite the presence of a genetic abnormality that might affect growth regulatory pathways in all cells, specific genetic abnormalities generally predispose only to certain tumor types. This selectivity highlights the observation that gene function contributes to growth or development only within a particular milieu or physiologic context.

Specific tumors occur earlier and are more often bilateral (in paired organs) when they result from germline mutations than when they result from sporadic or somatic alterations. Such is often the case in two pediatric malignancies, Wilms tumor and retinoblastoma. These observations led Alfred Knudson to propose a “two-hit” model of carcinogenesis in which the first genetic defect, already present in the germ line, must be complemented by an additional spontaneous mutation before a tumor can arise [62]. In sporadic cancer, cellular transformation occurs only when two (or more) spontaneous mutations take place in the same cell. The critical features of the Knudson model – the small number of mutations required for malignant transformation, the possible inheritance of a first mutation and the gradual disappearance of transformable target cells with increasing age, provide a conceptual framework for mutational theories of the genetics of most childhood tumors. In this scheme familial tumors will present earlier than sporadic tumors of the same histologic type; inheritance of a tumorigenic mutation will also predispose to multiple tumor occurrences.

Much more common, however, are somatically acquired chromosomal aberrations, which are confined to the malignant cells. These aberrations affect growth factors and their receptors, signal transducers, and transcription factors. The general types of chromosomal alterations associated with malignant transformation are shown in Fig. 3.1. Although a low level of chromosomal instability exists in a normal population of cells, neoplastic transformation occurs only if these alterations affect a growth-regulating pathway and confer a growth advantage.

Fig. 3.1
figure 1

The spectrum of gross chromosomal aberrations using chromosomes 1 and 14 as examples (Reprinted with permission from Look and Kirsch [76])

Full size image

DNA Content

Normal human cells contain two copies of each of 23 chromosomes; therefore, a normal “diploid” cell has 46 chromosomes. Although cellular DNA content, or ploidy, is accurately determined by karyotypic analysis, it can be estimated by the much simpler method of flow cytometry. The DNA index (DI) is defined as the ratio of the number of chromosome copies per cell to that of a normal cell (i.e., 46). Diploid cells have a DI of 1.0, whereas near-triploid cells have a DNA index ranging from 1.26 to 1.76. The majority (55 %) of primary neuroblastoma cells are triploid or near triploid, having between 58 and 80 chromosomes, whereas the remainder are near diploid (35–57 chromosomes) or near tetraploid (81–103 chromosomes) [57]. Neuroblastomas consisting of near-diploid or near-tetraploid cells usually have structural genetic abnormalities (e.g., chromosome 1p deletion and amplification of the MYCN oncogene), whereas those consisting of near-triploid cells are characterized by three almost complete haploid sets of chromosomes with few structural abnormalities [12]. The DI can have prognostic significance; patients with near-triploid tumors typically have favorable clinical and biologic prognostic factors, and excellent survival rates, compared with those who have near-diploid or near-tetraploid tumors [75].

Chromosomal Translocations

Many pediatric cancers, particularly soft-tissue neoplasms and hematologic malignancies, have recurrent, nonrandom abnormalities in chromosomal structure, typically chromosomal translocations (Table 3.2). The most common result of a nonrandom translocation is the fusion of two distinct genes from different chromosomes. The genes are typically fused within the reading frame and express a functional, chimeric protein product that has transcription factor or protein kinase activity. These fusion proteins contribute to tumorigenesis by activating genes or proteins involved in cell proliferation. For example, in Ewing sarcoma the consequence of the t(11;22)(q24;q12) translocation is the fusion of EWS, a transcription factor gene on chromosome 22, and FLI-1, a gene encoding a member of the ETS family of transcription factors on chromosome 11 [81]. The resultant chimeric protein, which contains the DNA-binding region of FLI-1 and the transcription activation region of EWS, has greater transcriptional activity than does EWS alone [82]. The EWS:FLI-1 fusion transcript is detectable in approximately 90 % of Ewing sarcomas. At least four other EWS fusions have been identified in Ewing sarcoma; fusion of EWS with ERG (another ETS family member) accounts for an additional 5 % of cases [124]. Alveolar rhabdomyosarcomas have characteristic translocations between the long arm of chromosome 2 (75 % of cases) or the short arm of chromosome 1 (10 % of cases) and the long arm of chromosome 13. These translocations result in the fusion of PAX3 (at 2q35) or PAX7 (at 1p36) with FOXO1, a gene encoding a member of the forkhead family of transcription factors [47]. The EWS:FLI-1 and PAX7:FOXO1 fusions appear to confer a better prognosis for patients with Ewing sarcoma and alveolar rhabdomyosarcoma, respectively [6, 32]. Translocations that generate chimeric proteins with increased transcriptional activity also characterize desmoplastic small round cell tumor [69], myxoid liposarcoma [108], extraskeletal myxoid chrondrosarcoma [23], malignant melanoma of soft parts [140], synovial sarcoma [24], congenital fibrosarcoma [130], cellular mesoblastic nephroma [111], and dermatofibrosarcoma protuberans [94].

Table 3.2 Common, recurrent translocations in soft tissue tumors
Full size table

Proto-oncogene Activation

Proto-oncogenes are commonly activated in transformed cells by gene amplification or point mutation. Gene amplification (i.e., selective replication of DNA sequences) enables a tumor cell to increase expression of crucial genes whose products are ordinarily tightly controlled. The amplified DNA sequences, or amplicons, may be maintained episomally (i.e., extrachromosomally) as double minutes—paired chromatin bodies lacking a centromere—or as intrachromosomal, homogeneously staining regions. In about one third of neuroblastomas, for example, the transcription factor and proto-oncogene MYCN is amplified. MYCN encodes a 64-kDa nuclear phosphoprotein (MycN) that forms a transcriptional complex by associating with other nuclear proteins expressed in the developing nervous system and other tissues [63]. Increased expression of MycN increases the rates of DNA synthesis and cell proliferation and shortens the G1 phase of the cell cycle [77]. The MYCN copy number in neuroblastoma cells can be amplified 5- and 500-fold and is usually consistent among primary and metastatic sites and at different times during tumor evolution and treatment [11]. This consistency suggests that MYCN amplification is an early event in the pathogenesis of neuroblastoma. Because MYCN gene amplification is usually associated with advanced stages of disease, rapid tumor progression, and poor outcome, it is a powerful prognostic indicator for neuroblastoma [13, 119]. The cell surface receptor gene ERBB2 is another proto-oncogene commonly overexpressed due to gene amplification, an event that occurs in breast cancer, osteosarcoma, and Wilms tumor [104].

An example of proto-oncogene activation by point mutation involves the tyrosine kinase receptor, anaplastic lymphoma kinase (ALK), on the short arm of chromosome 2 (2p23). Receptor tyrosine kinases (RTK) are high-affinity cell surface receptors for many growth factors, cytokines and hormones. When activated through ligand binding, these proteins mediate phosphorylation of tyrosine resides on target molecules or substrates, resulting in intracellular signaling and, ultimately, the regulation of normal cellular processes. Mutation of RTK’s can lead to constitutive activation of the signaling pathway in the absence of ligand. Recently, activating mutations of ALK have been shown to be the germline abnormality associated with hereditary neuroblastoma [89]. These mutations can also be somatically acquired, as can amplification of the gene, although the prevalence of ALK activation in sporadic neuroblastoma is not known [19]. Activated ALK has proven to be a targetable abnormality in neuroblastoma, with drugs such as crizotinib, an anti-ALK antibody, showing efficacy [21].

Another example is the RET proto-oncogene, which encodes a RTK for members of the glial cell line-dervied neurotrophic factor family of extracellular signaling molecules. Mutation of the RET proto-oncogene, resulting in gain of function, is associated with medullary thyroid carcinoma, pheochromocytoma and multiple endocrine neoplasias types 2A and 2B [91]. Interestingly, the various specific mutations have a different influence on the tumor phenotype [90].

Inactivation of Tumor Suppressor Genes

Tumor suppressor genes, or antioncogenes, provide negative control of cell proliferation. Loss of function of the proteins encoded by these genes, through deletion or mutational inactivation of the gene, liberates the cell from growth constraints and contributes to malignant transformation. The cumulative effect of genetic lesions that activate proto-oncogenes or inactivate tumor suppressor genes is a breakdown in the balance between cell proliferation and cell loss due to differentiation or apoptosis. Such imbalance results in clonal overgrowth of a specific cell lineage. The first tumor suppressor gene to be recognized was the retinoblastoma susceptibility gene, RB. This gene encodes a nuclear phosphoprotein that acts as a “gatekeeper” of the cell cycle. RB normally permits cell cycle progression through the G1 phase when it is phosphorylated but prevents cell division when it is unphosphorylated. Inactivating deletions or point mutations of RB cause the protein to lose its regulatory capacity.

The nuclear phosphoprotein, p53, has also become recognized as an important tumor suppressor gene, perhaps the most commonly altered gene in all human cancers. Inactivating mutations of p53 also cause the protein to lose its ability to regulate the cell cycle. The p53 gene is frequently inactivated in solid tumors of childhood, including osteosarcoma, rhabdomyosarcoma, brain tumors, anaplastic Wilms tumor, and a subset of chemotherapy-resistant neuroblastoma [5, 59, 66]. In addition, heritable cancer-associated changes in the p53 tumor suppressor gene occur in families with Li-Fraumeni syndrome, an autosomal dominant predisposition for rhabdomyosarcoma, other soft tissue and bone sarcomas, premenopausal breast cancer, brain tumors, and adrenocortical carcinomas [78].

Recently, inactivating mutations of ATRX, a transcriptional regulator that is part of a multiprotein complex that plays a role in regulating chromatin remodeling, nucleosome assembly, and telomere maintenance, have been found in neuroblastoma, particularly high stage tumors in older patients [22]. ATRX mutations appear to be loss-of-function mutations associated with an absence of the ATRX protein in the nucleus, and with long telomeres. How these alterations lead to lengthened telomeres is uncertain, however. These results may provide a molecular marker and potential therapeutic target for neuroblastoma among adolescents and young adults. It may also delineate the subset of children with neuroblastoma who have a chronic but progressive clinical course when receiving standard therapeutic approaches who may require a different treatment strategy.

Other tumor suppressor genes inactivated in pediatric malignancies include Wilms tumor 1 (WT1), neurofibromatosis 1 (NF1), and von Hippel-Lindau (VHL). In addition, other tumor suppressor genes are presumed to exist but have not been definitively identified. For example, early karyotype analyses of neuroblastoma-derived cell lines found frequent deletion of the short arm of chromosome 1 [14]. Deletion of genetic material in tumors suggests the presence (and subsequent loss) of a tumor suppressor gene, but no individual tumor suppressor gene has been identified on chromosome 1p. Functional confirmation of the presence of a 1p tumor suppressor gene comes from the demonstration that transfection of chromosome 1p into a neuroblastoma cell line results in morphologic changes of differentiation and ultimately cell senescence [4]. Approximately 20–35 % of primary neuroblastomas exhibit 1p deletion, as determined by fluorescent in situ hybridization (FISH), and the smallest common region of loss is located within region 1p36 [45]. Deletion of 1p is also common in Wilms tumor [48]. Other chromosomal regions, whose loss in tumor cells suggests the loss of a tumor suppressor gene, include 11q in neuroblastoma [3] and 16q in Wilms tumor [49].

Genetic Variants

Cancer epidemiology genome-wide association studies (GWAS) examine the association of common genetic variants, most often single nucleotide polymorphisms (SNPs) or copy number variations (CNV), with the presence of a particular cancer. The causal relationship between the DNA variant associated with the cancer is not always certain but an excessive inheritance of “risk” variants has been postulated to increase susceptibility to the disease. Several GWAS studies have been performed in patients with neuroblastoma and Wilms tumors and have identified a number of such genetic risk variants [80, 129]. These observations suggest that developmental childhood cancers are likely influenced by common DNA variations, leading to the development of a putative genetic model (Fig. 3.2) [80]. Recent data suggest that the higher prevalence of high-risk disease in Black and Native American patients with neuroblastoma may be associated with certain genetic variants found more commonly in these ethnic groups [52].

Fig. 3.2
figure 2

A threshold for the development of neuroblastoma may exist which can be reached through a combination of inherited genetic factors together with the effects of environmental exposures. Certain mutations (e.g., to ALK or PHOX2B) may themselves allow a cell to reach this threshold. Alternatively, common (but less impactful) DNA variations in a larger number of genes such as FLJ22536, BARD1, and NBPF23, may combine to allow a cell to reach this threshold (Reprinted with permission from Maris [80])

Full size image

Finally, GWAS studies have revealed genome variations that affect not only susceptibility to specific cancers but that may also influence the pharmacokinetic and pharmacodynamic characteristics of administered chemotherapeutics. Several genetic factors with relatively small effects may combine in the determination of a pharmacogenomic phenotype [102]. Because the therapeutic index of many drugs, especially in children, is very narrow with substantial risk for toxicity at doses required for therapeutic effects, identifying and understanding the impact of these variants is critical for optimizing treatment.

Epigenetics

As stated previously, the hallmark of cancer is dysregulated gene expression. However, not only do genetic factors influence gene expression but epigenetic factors do as well, with these factors being at least important as genetic changes in their contribution to the pathogenesis of cancer. Epigenetic alterations are defined as those heritable changes in gene expression that do not result from direct changes in DNA sequence. Mechanisms of epigenetic regulation most commonly include DNA methylation and modification of histones, although the contribution of microRNAs (miRNA), a class of noncoding RNAs, is becoming increasingly recognized. Whole-genome sequencing of tumors, made possible recently by significant advances in technology, has been performed to investigate the genetic landscape of a variety of pediatric tumors [36]. Initially it was felt that early alterations of genes such as RB and MYCN may underlie the rapid acquisition of cooperating mutations in key cancer pathways through chromosome instability. However, few recurring amino acid changes have been detected in retinoblastoma and neuroblastoma specimens [22, 88, 138], suggesting that the tumor genomes are more stable than previously believed. However, unlike the genetic landscape, the epigenetic profiles showed profound changes suggesting that epigenetic changes may have a more dominant role in pediatric tumorigenesis.

DNA methylation is a reversible process that involves methylation of the fifth position of cytosine within CpG dinucleotides present in DNA. These dinculeotides are usually in the promoter regions of genes; methylation of these sites typically causes gene silencing, thereby preventing expression of the encoded proteins. This process is part of the normal mechanism for imprinting, X-chromosome inactivation and generally keeping large areas of genomic DNA silent, but may also contribute to the pathogenesis of cancer by silencing tumor suppressor genes. However, both abnormal hypo- and hypermethylation states exist in human tumors, resulting in both dysregulated expression and silencing, respectively, of affected genes. These modifications of the nucleotide backbone of human DNA are becoming increasingly recognized in human cancer both for their frequency and importance. For example, promoter methylation resulting in silencing of caspase 8, a protein involved in apoptosis, likely contributes to the pathogenesis of MYCN-amplified neuroblastoma [127], as well as Ewing sarcoma [46].

Histones are the proteins that give structure to DNA, and together with the DNA form the major components of chromatin. The functions of histones are to package DNA into a smaller volume to fit in the cell, to strengthen the DNA to allow replication, and to serve as a mechanism to control gene expression. Alterations in histones can mediate changes in chromatin structure. The compacted form of DNA, termed heterochromatin, is largely inaccessible to transcription factors and, therefore, genes in the affected regions are silent. Other modifications of histones can cause DNA to take a more open or extended configuration (euchromatin), allowing for gene transcription. The N-terminal tails of histones can be modified by a number of different processes including methylation and acetylation, mediated by histone acetyl transferases (HAT) and deacetylases (HDAC), and histone methyltransferases (HMT). Each of these processes alters histone function, which, in turn alters the structure of chromatin and, therefore, the accessibility of DNA to transcription factors. Methylation of the DNA itself can also effect changes in chromatin structure.

MicroRNAs are a group of small, regulatory noncoding RNAs that appear to function in gene regulation. These miRNAs are single–stranded RNA fragments of 21–23 nucleotides that are complementary to encoding mRNAs [113]. Their function is to down-regulate expression of target mRNAs; it is estimated that miRNAs regulate the expression of about 30 % of all human genes [73]. These miRNAs regulate gene expression primarily by incorporating into silencing machinery called RNA-induced silencing complexes (RISC). MiRNAs are involved in a number of fundamental biologic processes, including development, differentiation, cell cycle regulation and senescence. However, broad analyses of miRNA expression levels has demonstrated that many miRNAs are dysregulated in a variety of different cancer types, including neuroblastoma and other pediatric tumors [131], frequently losing their function as gene silencers/tumor suppressors. The activity of miRNAs, like gene expression, is also under epigenetic regulation.

Metastasis

Metastasis is the spread of cancer cells from a primary tumor to distant sites and is the hallmark of malignancy. The development of tumor metastases is the main cause of treatment failure and a significant contributing factor to morbidity and mortality resulting from cancer. Although the dissemination of tumor cells through the circulation is probably a frequent occurrence, the establishment of metastatic disease is a very inefficient process. It requires several events, including entry of the neoplastic cells into the blood or lymphatic system; survival in the circulation, avoidance of immune surveillance, invasion of foreign (heterotopic) tissues, and the establishment of a blood supply to permit expansion of the tumor at the distant site. Simple, dysregulated cell growth is not sufficient for tumor invasion and metastasis. Many tumors progress through distinct stages that can be identified by histopathologic examination, including hyperplasia, dysplasia, carcinoma in situ, invasive cancer, and disseminated cancer. Genetic analysis of these different stages of tumor progression suggests that uncontrolled growth results from progressive alteration in cellular oncogenes and inactivation of tumor suppressor genes but these genetic changes driving tumorigenicity are clearly distinct from those that determine the metastatic phenotype.

Histologically, invasive carcinoma is characterized by lack of a basement membrane around an expanding mass of tumor cells. Matrix proteolysis appears to be a key part of the mechanism of invasion by tumor cells, which must be able to move through connective tissue barriers, such as the basement membrane, to spread from their site of origin. The proteases involved in this process include matrix metalloproteinases and their inhibitors, tissue inhibitors of matrix metalloproteinases. The local environment of the target organ may profoundly influence the growth potential of extravasated tumor cells [39]. The various cell-surface receptors that mediate interactions between tumor cells and between tumor cells and the extracellular matrix include cadherins, integrins (transmembrane proteins formed by the noncovalent association of α and β subunits), and CD44, a transmembrane glycoprotein involved in cell adhesion to hyaluronan [128]. Tumor cells must decrease their adhesiveness to escape from the primary tumor, but at later stages in metastasis, the same tumor cells need to increase their adhesiveness during arrest and intravasation to distant sites.

Angiogenesis

Angiogenesis is the biologic process of new blood vessel formation. This complex, invasive process involves multiple steps including proteolytic degradation of the extracellular matrix surrounding existing blood vessels, chemotactic migration and proliferation of endothelial cells, the organization of these endothelial cells into tubules, the establishment of a lumen that serves as a conduit between the circulation and an expanding mass of tumor cells, and functional maturation of the newly formed blood vessel [44, 110]. Angiogenesis involves the coordinated activity of a wide variety of molecules including growth factors, extracellular matrix proteins, adhesion receptors, and proteolytic enzymes. Under physiologic conditions the vascular endothelium is quiescent and has a very low rate of cell division, such that only 0.01 % of endothelial cells are dividing [44, 53, 110]. However, in response to hormonal cues or hypoxic or ischemic conditions, the endothelial cells can be activated to migrate, proliferate rapidly, and create lumens.

Angiogenesis occurs as part of such normal physiologic activities as wound healing, inflammation, the female reproductive cycle, and embryonic development. In these processes, angiogenesis is tightly and predictably regulated. However, angiogenesis can also be involved in the progression of several pathologic processes in which there is a loss of regulatory control that results in persistent growth of new blood vessels. Such unabated neovascularization occurs in rheumatoid arthritis, inflammatory bowel disease, hemangiomas of childhood, ocular neovascularization, and the growth and spread of tumors [43].

Compelling data implicate the requirement for tumor-associated neovascularization in tumor growth, invasion, and metastasis [9, 41, 42, 105]. A tumor in the prevascular phase (i.e., before new blood vessels have developed) can grow to only a limited size, approximately 2–3 mm3. At this point the rapid cell proliferation is balanced by equally rapid cell death by apoptosis, resulting in a nonexpanding tumor mass. The switch to an angiogenic phenotype with tumor neovascularization results in a decrease in the rate of apoptosis, thereby shifting the balance to cell proliferation and tumor growth [55, 74]. This decrease in apoptosis occurs, in part, because the increased perfusion resulting from neovascularization permits improved nutrient and metabolite exchange. In addition, the proliferating endothelium may supply, in a paracrine manner, a variety of factors that promote tumor growth, such as insulin-like growth factors I and II (IGF-I, IGF-II) [51].

In experimental models, increased tumor vascularization correlates with increased tumor growth, whereas restriction of neovascularization limits tumor growth. Clinically, the onset of neovascularization in many human tumors is temporally associated with increased tumor growth [126], and high levels of angiogenic factors are commonly detected in blood and urine from patients with advanced malignancies [93]. In addition, the number and density of new microvessels within primary tumors have been shown to correlate with the likelihood of metastasis as well as the overall prognosis for patients with a wide variety of neoplasms, including pediatric tumors such as neuroblastoma and Wilms tumor [1, 83].

It has become increasingly evident that the regulation of tumor angiogenesis is complex: new blood vessel formation occurs as the result of competing pro- and antiangiogenic signals originating in multiple tissues [20]. Specific genetic events in certain cancers, such as altered expression of the p53 tumor suppressor gene [28, 139] or the human EGFR gene [71, 103, 136], not only affect the cell cycle but also play a role in angiogenesis by modulating key signals (e.g., upregulating the expression of vascular endothelial growth factor [VEGF], or downregulating the expression of the endogenous angiogenesis inhibitor thrombospondin 1).

Metastasis also appears to be dependent on angiogenesis [40, 74]. This dependence is probably due to several factors. First, new blood vessels in the primary tumor provide increased opportunities for the shedding of tumor cells into the circulation. Also, disruption of the basement membrane by proteases released by the proliferating endothelial cells may contribute to the metastatic potential of a tumor [15, 112]. Finally, successful growth of metastatic cells in foreign target organs depends on the stimulation and formation of new blood vessels, perhaps even when cells metastasize to the bone marrow.

Environmental Carcinogenesis

As stated previously, tumorigenesis is a complex process in which the progressive acquisition of combinations of critical genetic and epigenetic alterations shifts normal cells into uncontrolled growth and clonal expansion. These alterations in the genome can either be inherited or acquired, the later being as a result of the influence of factors intrinsic to a dividing cell or of the environment on the host genome. The most significant host factor relates to the normal, albeit, low rate of inaccurate DNA replication that goes uncorrected by normal host mechanisms. Living organisms have been selected for their ability to accurately replicate their genomes, although not with absolute precision. This ensures stability while permitting the dynamic of genetic change essential to environmental adaptation and consequent evolution. Variability in the local environment may make such mistakes more or less likely. This rate of accumulating alterations in the genome can also be significantly increased when there are defects in the normal genetic corrective mechanisms, or when there is increased host cell genomic instability. Critical for tumorigenesis, however, in addition to the initial occurrence of DNA damage, is the persistence of the DNA alterations and their eventual transmission to clonal descendants of the originally affected cell. Two conditions need to be fulfilled for persistence and inheritance of DNA damage. (1) The DNA repair systems of the cell fail to remove or correct the damage and (2) the residual lesion should not only be compatible with continued cell viability and proliferative capacity but also should confer a survival advantage.

The first line of cellular defense against DNA damage is the recognition of the damage and the implementation of a variety of molecular mechanisms which have evolved to effect repair. An important class of genetic lesions are mismatches which result in non-complementary DNA sequence over a short region. Unrepaired mismatches generate point mutations at the next round of cell division and the synthesis of mutant or truncated proteins. Removal of these mismatches and restoration of normal complementary base-pairing is the responsibility of mismatch repair enzymes. Other DNA lesions require different sorts of repair. DNA strands can be broken, generating a spectrum of lesions from point mutations to large scale chromosomal aberrations. Several mechanisms exist for strand break repair including the process of homologous recombination. The DNA repair processes are intimately linked to cell cycle control in proliferating cells, with several check points existing at which DNA-damaged cells are blocked until the repair processes have been completed. One example is the p53 tumor suppressor gene and its role in cell cycle arrest. DNA damage by a variety of extrinsic agents leads to cellular accumulation of the p53 protein, largely by stabilization of the protein, and the resultant blocking of the damaged cell at the G1 checkpoint. Successful repair of DNA damage leads to a reduction in p53 levels and release from the G1 block. However, incomplete or unsuccessful repair continues to generate the p53 blocking signal. Long-term-G1 blocking then invokes a cell death pathway, usually apoptosis, to eliminate unrepairable mutant cells. Thus p53 plays a critical role as the “guardian of the genome” to prevent the onset of genetic instability [70].

Epidemiology

The opportunity to practice prevention of cancer depends on the existence of potentially avoidable factors and their recognition so that appropriate action can be taken. The impact or causal role of environmental factors on the development of human malignancies was first recognized by noting unexpectedly high cancer incidences in certain occupational groups. Fabia and Thuy first suggested that a father’s occupation might increase the risk of a child developing cancer [37]. The ability of certain chemicals was then documented in various animal models of carcinogenesis. In addition, population based studies confirmed histology and anatomic site-specific cancer rates among geographically distinct populations. Changes in cancer frequency among migrating ethnic groups, high cancer rates associated with specific occupations and most notably the risk of cancer associated with such activities as smoking and tobacco use have confirmed that environmental and lifestyle exposures contribute to human cancer risk. In addition, it has become recognized that certain people carry hereditary susceptibility genes that increase risk for developing cancer with particular environmental exposures. Environmental agents may cause mutations which are distinct or different from the predominant mutational type resulting from intrinsic mutagenesis or from the action of other environmental agents. This gives rise to the possibility of “molecular fingerprinting” by which environmental agents might be identified by the characteristic mutational type they have caused in the oncogenes or tumor suppressor genes of a tumor suspected of having environmental causation. An example is the characteristic mutations in the p53 gene associated with aflatoxin-mediated hepatocellular carcinoma [56].

The critical corollary to the identification of these factors is that exposure or lifestyle modification may be able to decrease the incidence of cancer development. Epidemiology, broadly defined, is the study of disease occurrence in different population groups in order to help identify causative risk factors and to plan appropriate preventive strategies. Epidemiology may also provide clues to etiology and pathogenesis.

Strong evidence exists that a substantial proportion of adult cancers are environmentally influenced, with tobacco, alcohol and diet being among the most important factors [2]. It is much less obvious that environmental factors play an important role in the development of childhood cancer. This is likely due to the fundamental differences between oncogenesis in children and adults that influence the impact of environmental exposures on carcinogenesis. Most adult tumors are of epithelial origin, such as in the gastrointestinal or respiratory tracts, surfaces that are directly exposed to carcinogens, whereas most pediatric tumors are of mesenchymal origin in tissues or sites with minimal contact with the environment. In addition, during embryogenesis fetal tissues are normally undergoing rapid cell division with high rates of proliferation, much like cancer cells, and then ultimately undergo differentiation or apoptosis. Additionally, adult cancers, despite variations between pathologic types, usually conform to a pattern in which the incidence increases with age, reflecting an accumulation of multiple mutations with time. Childhood cancers, in distinction to those in adults, typically have an incidence that initially increases with age, reaches a peak and then falls (Fig. 3.3), suggesting that other epigenetic factors, other than simply an accumulation of genetic mutations contribute to the development of malignancies in children. Finally, the cumulative effects of carcinogens such as irradiation and other environmental exposures often are not apparent for many years and, therefore, are less likely to have a direct impact on the development of pediatric malignancies. Thus, few environmental factors have been identified as being associated with pediatric oncogenesis. Nevertheless some environmental factors have been associated with the evolution of pediatric tumors and are discussed below.

Fig. 3.3
figure 3

Schematic illustration of the differing age-incidence patterns of adult and childhood cancer (not to scale)

Full size image

Potential Causative Agents

Radiation

Ionizing radiation is tumorigenic and is capable of causing or contributing to the development of a wide variety of malignancies. It causes a variety of heritable DNA lesions, from point mutations to chromosomal deletions or rearrangements, induces transformation to a malignant state in cells in culture and causes a range of types of cancer in experimental animals in a dose-dependent fashion. The developing fetus is particularly sensitive to the effects of ionizing irradiation, which increases the risk of childhood cancer by approximately 6 %/Gy exposure [34]. For postnatal irradiation the risk is about half of that for the fetus [109]. The radiation being delivered to children as part of diagnostic imaging studies, particularly CT scans, is currently being closely scrutinized as its potential role in the subsequent development of cancer is becoming increasingly appreciated [35]. It should be noted, however, that most of the cancers to which childhood irradiation makes a contribution will appear in adulthood, as radiation-induced cancer generally has a very long latency period [86]. Non-ionizing radiation such as ultraviolet light and electromagnetic fields may also contribute to the development of cancer, although, again, the association of skin cancer with UV exposure, for example, is well known but typically results in a heightened incidence in adult years. The epidemiologic evidence that electromagnetic radiation leads to cancer is conflicting [68], and the data regarding the effects of high-voltage transmission lines, electrical appliances and video display screens do not seem to support a causal role for these factors.

Chemical Agents

There are many chemical agents which have DNA-damaging capacity and therefore tumorigenic potential, but few for which there is any clear evidence of significant involvement in the causation of childhood cancer. Several chemical factors have been suggested as relevant to childhood cancer, including pesticides, vitamin K administration, passive cigarette smoke and maternal use of “recreational” drugs. An uncommon but very striking factor is the well-known risk associated with use of the hormone diethylstilbestrol which used to be administered in pregnancy in some cases where miscarriage was anticipated. This resulted in a 0.1 % risk of the development of clear cell adenocarcinomas of the cervix or vagina in female offspring [87]. There have been some reports of other childhood cancers, particularly neuroblastoma, associated with induced ovulation, although the evidence is not overwhelming [79].

Viruses

The role played by several viruses in human malignancies has been well established, most notably Epstein-Barr virus (EBV) and its association with Burkitt lymphoma, whereby immortalization of B cells by EBV has been suggested to be the initial event in multistep carcinogenesis [61]. EBV also appears to have a causal role in the development of nasopharyngeal carcinoma and some cases of Hodgkin lymphoma [50]. Cancer also occurs with increased frequency in children with human immunodeficiency virus (HIV) infection. The most common types are lymphomas and leukemias, although solid tumors including leiomyosarcoma and Kaposi sarcoma also occur with an increased incidence [106, 123]. However, neither the specific clinical, immunological, and viral risk factors for malignancy in these patients, nor the pathogenesis of HIV-related pediatric malignancies have been clearly elucidated.

Parental Occupation and Exposure to Noxious Agents

Parental occupation and exposures have been linked to an increased risk of a variety of childhood cancers [26, 116]. Transgenerational effects may be due to direct germ cell mutation, transport of carcinogens in the semen or epigenetic alterations of gene expression [100]. Paternal exposure to solvents such as benzene, xylene, toluene and carbon tetrachloride have been implicated in the pathogenesis of hematologic malignancies and brain tumors as have paints and pesticides [95]. Increased risk of childhood cancers such as leukemia, Ewing sarcoma, Germ cell tumor and Wilms tumor have been associated with certain paternal occupations including auto mechanic, welder, and to exposure at work to motor vehicle exhaust fumes, pesticides, petroleum and ether [54, 120, 122]. Although increased risk of some childhood cancers in association with potential carcinogen exposure is suggested by multiple studies, methodological limitations common to many studies restrict conclusions; these include exposure classification, small sample size and potential biases in control selection.

Iatrogenic Factors

Because the survival rates for childhood cancers have improved to more than 80 %, the proportion of childhood cancer survivors within the general population increases every year. Survivors are at risk for multiple late sequelae of therapy, including the development of a secondary malignancy. A significant factor contributing to this risk, in addition to the genetic predisposition of the patient, is the type of therapy received [7]. For example, an increased risk of subsequent leukemia is well-documented after exposure to epipodophyllotoxins and alkylating agents [107]. Similarly, the risk of carcinomas of the breast and thyroid, particularly after treatment for childhood Hodgkin lymphoma, has been extensively reported, and is related, in part to exposure to ionizing radiation as part of the treatment of the initial cancer [8]. Other examples include therapy-related brain tumors after cranial irradiation, osteosarcoma after irradiation for retinoblastoma and tumors, such as thyroid cancer, arising as a complication of low-dose irradiation given in the past as treatment for tinea capitis and acne. These secondary malignancies often occur in adulthood but may occur late in the teenage years.

Tumor Types

Neuroblastoma

Several case-control studies have examined the relationship between maternal and paternal occupation and exposure, and the risk of neuroblastoma in offspring [18, 84, 125, 134]. Two studies found an association with fathers employed in electronics-related occupations including electricians and welders (odds ratio = 11.7, 95 % confidence interval 1.4–98.5) [125]. Another study found increased risks in electrical, farming and gardening, and painting occupations [98]. A variety of other paternal occupations and industries have been shown to have an increased risk of having a child with neuroblastoma [18, 134]. Paternal exposures to hydrocarbons such as diesel fuel, lacquer thinner and turpentine were associated with an increased incidence of neuroblastoma as were exposures to wood dust and solders [33]. Pesticide use in both home and garden were modestly associated with neuroblastoma [29]. Certain maternal occupations have also been found to have an association with an increased risk [18, 125, 134].

Several epidemiologic and case series have suggested a relationship between the use of certain medications just prior to and during pregnancy and neuroblastoma, specifically hormone use and fertility drugs [64, 85, 118]. although others studies have not confirmed such an association [27]. One study by Schuz et al. observed a positive association with the use of oral contraceptives or other sex hormones during pregnancy (particularly with male offspring), a shorter gestational duration, lower birth weight, and maternal alcohol consumption during pregnancy [117]. Other drugs have been implicated although the data have not always been consistent among different studies. Similarly the results for smoking, alcohol use and the use of hair dye in some studies are suggestive but not conclusive [60, 64, 118]. while other studies find no association [137]. Maternal use of any illicit or recreational drug around pregnancy has been associated with an increased risk of neuroblastoma in offspring (odds ratio = 1.82, 95 % confidence interval 1.13–3.00), particularly the use of marijuana in the first trimester of pregnancy [10]. Other studies have suggested an association between maternal hair dye use and elevated risk of childhood cancer including neuroblastoma (OR = 1.6, 95 %CI = 1.2–2.0). Vitamin use during pregnancy might reduce the incidence of neuroblastoma, consistent with findings for other childhood cancers [99]. Also, children with neuroblastoma were less likely to have breast-fed than control children (CR = 0.6, 95 % CI = 0.5–0.9) with the decreased association between breast-feeding and neuroblastoma increasing with increasing duration of breast-feeding.

Wilms Tumor

The first suggestion that paternal occupational exposures might be of importance in the etiology of Wilms tumors came from Kantor et al. [58] From a comparison of birth certificates for 149 Connecticut tumor registry cases with 149 matched controls, they estimated relative risks of 2.4 for hydrocarbon-related occupations and 3.7 for those with a potential for lead exposure. This was later supported by Wilkins and Sinks although their results did not reach statistical significance for the association [135]. Others have attempted to confirm this finding [17, 67, 114]. Although suggestive associations have been found in some studies for machinists, mechanics and welders the numbers are small and the patterns are inconsistent. Olshan et al. found no consistent pattern of increased risk for paternal exposure to hydrocarbons and lead but did find that certain paternal occupations did have an elevated odds ratio of Wilms tumor including vehicle mechanics, auto body repairmen and welders [96]. Offspring of fathers who were auto mechanics had a 4- to 7-fold increased risk of Wilms tumor. Other early studies have suggested possible associations with maternal smoking, coffee/tea drinking and exposure to synthetic progestins during pregnancy and the use of hair coloring products [17, 72, 95]. However, these studies are subject to several methodologic limitations including misclassification of exposure, selection bias, and small sample size and later studies have, in general, failed to confirm most of the previously reported maternal risk factors for Wilms tumor [97]. Breast feeding was associated with a reduced risk of Wilms tumor (odds ratio = 0.7, 95 % confidence interval = 0.5–0.9).

Liver Tumors

Environmental factors have also been implicated in hepatoblastoma. An association with certain occupational exposures in fathers of children with hepatoblastoma, including excess exposures to metals such as in welding and soldering fumes (odds ratio 8.0), petroleum products, and paints (odds ratio 3.7), has been observed [16]. Prenatal exposure to acetaminophen in combination with petroleum products has also been noted in association with hepatoblastoma [115]. There is a striking association of hepatoblastoma with prematurity, with the relative risk increasing with decreasing birth weight [38]. However, the etiology behind this association is currently unknown. An increased incidence of liver tumors is also seen in association with fetal alcohol syndrome, exposure to hepatitis B and aflatoxin, and prolonged parenteral nutrition in infancy [101]. The most striking association, however, is in children with metabolic diseases such as tyrosinemia. In children with these disorders, the tissues are exposed to high, continuous levels of endogenous carcinogens, and are at such high risk for the development of malignancies such as hepatocellular carcinoma early in life, that early organ transplantation is recommended [25, 132].

Summary

The potential role of environmental exposures in the etiology of childhood cancer remains uncertain. The relatively few epidemiologic studies that have been conducted have been limited by a number of confounding factors, including sample size, exposure misclassification and selection bias. Nevertheless, sufficient suggestive data exist to warrant further evaluation into the role of environmental exposures in pediatric oncogenesis. The goal being, of course, to identify factors that can be eliminated or avoided in order to decrease the risk for developing a malignancy.