Pediatric Diseases

Etiology

Known causes of errors in human malformations can be grouped into three major categories: genetic, environmental, and multifactorial (Table 6–7). The cause has not been identified for almost half of the reported cases.

Table 6–7 Causes of Congenital Malformations in Humans

* Live births.

Data from Stevenson RE, Hall JG, Goodman RM (eds): Human Malformations and Related Anomalies. New York, Oxford University Press, 1993, p 115.

Genetic causes of malformations include all of the previously discussed mechanisms of genetic disease. Virtually all chromosomal syndromes are associated with congenital malformations. Examples are Down syndrome and other trisomies, Turner syndrome, and Klinefelter syndrome. Most chromosomal disorders arise during gametogenesis and hence are not familial. Single-gene mutations, characterized by mendelian inheritance, may underlie major malformations. For example, holoprosencephaly is the most common developmental defect of the forebrain and midface in humans (see Chapter 22); the Hedgehog signaling pathway plays a critical role in the morphogenesis of these structures, and loss-of-function mutations of individual components within this pathway are reported in families with a history of recurrent holoprosencephaly.

Environmental influences, such as viral infections, drugs, and radiation to which the mother was exposed during pregnancy, may cause fetal malformations (the appellation of “malformation” is used loosely in this context, since technically, these anomalies represent disruptions). Among the viral infections listed in Table 6–7, rubella was a major scourge of the 19th and early 20th centuries. Fortunately, maternal rubella and the resultant rubella embryopathy have been virtually eliminated in developed countries as a result of vaccination. A variety of drugs and chemicals have been suspected to be teratogenic, but perhaps less than 1% of congenital malformations are caused by these agents. The list includes thalidomide, alcohol, anticonvulsants, warfarin (oral anticoagulant), and 13-cis-retinoic acid, which is used in the treatment of severe acne. For example, thalidomide, once used as a tranquilizer in Europe and currently used for treatment of certain cancers, causes an extremely high incidence (50% to 80%) of limb malformations. Alcohol, perhaps the most widely used agent today, is an important environmental teratogen. Affected infants show prenatal and postnatal growth retardation, facial anomalies (microcephaly, short palpebral fissures, maxillary hypoplasia), and psychomotor disturbances. These features in combination are designated the fetal alcohol syndrome. While cigarette smoke–derived nicotine has not been convincingly demonstrated to be a teratogen, there is a high incidence of spontaneous abortions, premature labor, and placental abnormalities among pregnant smokers; babies born to mothers who smoke often have a low birth weight and may be prone to the sudden infant death syndrome (SIDS). In light of these findings, it is best to avoid nicotine exposure altogether during pregnancy. Among maternal conditions listed in Table 6–7, diabetes mellitus is a common entity, and despite advances in antenatal obstetric monitoring and glucose control, the incidence of major malformations in infants of diabetic mothers stands between 6% and 10% in most reported series. Maternal hyperglycemia–induced fetal hyperinsulinemia results in fetal macrosomia (organomegaly and increased body fat and muscle mass); cardiac anomalies, neural tube defects, and other CNS malformations are some of the major anomalies seen in diabetic embryopathy.

Multifactorial inheritance, which implies the interaction of environmental influences with two or more genes of small effect, is the most common genetic cause of congenital malformations. Included in this category are some relatively common malformations such as cleft lip and palate and neural tube defects. The importance of environmental contributions to multifactorial inheritance is underscored by the dramatic reduction in the incidence of neural tube defects by periconceptional intake of folic acid in the diet. The recurrence risks and mode of transmission of multifactorial disorders are described earlier in this chapter.

Pathogenesis

The pathogenesis of congenital anomalies is complex and still poorly understood, but two general principles of developmental pathology are relevant regardless of the etiologic agent:

1 The timing of the prenatal teratogenic insult has an important impact on the occurrence and the type of anomaly produced. The intrauterine development of humans can be divided into two phases: (1) the embryonic period, occupying the first 9 weeks of pregnancy, and (2) the fetal period, terminating at birth.

In the early embryonic period (first 3 weeks after fertilization), an injurious agent damages either enough cells to cause death and abortion or only a few cells, presumably allowing the embryo to recover without developing defects. Between the third and the ninth weeks, the embryo is extremely susceptible to teratogenesis, and the peak sensitivity during this period occurs between the fourth and the fifth weeks. During this period organs are being crafted out of the germ cell layers.

The fetal period that follows organogenesis is marked chiefly by the further growth and maturation of the organs, with greatly reduced susceptibility to teratogenic agents. Instead, the fetus is susceptible to growth retardation or injury to already formed organs. It is therefore possible for a given agent to produce different anomalies if exposure occurs at different times of gestation.

2 The complex interplay between environmental teratogens and intrinsic genetic defects is exemplified by the fact that features of dysmorphogenesis caused by environmental insults often can be recapitulated by genetic defects in the pathways targeted by these teratogens. Some representative examples follow:

Cyclopamine is a plant teratogen. Pregnant sheep who feed on plants containing cyclopamine give birth to lambs that have severe craniofacial abnormalities including holoprosencephaly and cyclopia (single fused eye—hence the origin of the moniker cyclopamine). This compound is a potent inhibitor of Hedgehog signaling in the embryo, and as stated previously, mutations of Hedgehog genes are present in subsets of fetuses with holoprosencephaly.

Valproic acid is an antiepileptic and a recognized teratogen. Valproic acid disrupts expression of a family of highly conserved developmentally critical transcription factors known as homeobox (HOX) proteins. In vertebrates, HOX proteins have been implicated in the patterning of limbs, vertebrae, and craniofacial structures. Not surprisingly, mutations in HOX family genes are responsible for congenital anomalies that mimic features observed in valproic acid embryopathy.

The vitamin A (retinol) derivative all-trans-retinoic acid is essential for normal development and differentiation, and its absence during embryogenesis results in a constellation of malformations affecting multiple organ systems, including the eyes, genitourinary system, cardiovascular system, diaphragm, and lungs (see Chapter 7 for vitamin A deficiency in the postnatal period). Conversely, excessive exposure to retinoic acid also is teratogenic. Infants born to mothers treated with retinoic acid for severe acne have a predictable phenotype (retinoic acid embryopathy), including CNS, cardiac, and craniofacial defects, such as cleft lip and cleft palate. The last entity may stem from retinoic acid–mediated deregulation of components of the transforming growth factor-β (TGF-β) signaling pathway, which is involved in palatogenesis. Mice with knockout of the Tgfb3 gene uniformly develop cleft palate, once again underscoring the functional relationship between teratogenic exposure and signaling pathways in the causation of congenital anomalies.

Immune Hydrops

Immune hydrops results from an antibody-induced hemolytic disease in the newborn that is caused by blood group incompatibility between mother and fetus. Such an incompatibility occurs only when the fetus inherits red cell antigenic determinants from the father that are foreign to the mother. The most common antigens to result in clinically significant hemolysis are the Rh and ABO blood group antigens. Of the numerous antigens included in the Rh system, only the D antigen is a major cause of Rh incompatibility. Fetal red cells may reach the maternal circulation during the last trimester of pregnancy, when the cytotrophoblast is no longer present as a barrier, or during childbirth itself (fetomaternal bleed). The mother then becomes sensitized to the foreign antigen and produces antibodies that can freely traverse the placenta to the fetus, in which they cause red cell destruction. With initiation of immune hemolysis, progressive anemia in the fetus leads to tissue ischemia, intrauterine cardiac failure, and peripheral pooling of fluid (edema). As discussed later, cardiac failure may be the final pathway by which edema occurs in many cases of nonimmune hydrops as well.

Several factors influence the immune response to Rh-positive fetal red cells that reach the maternal circulation:

Appreciation of the role of previous sensitization in the pathogenesis of Rh-hemolytic disease of the newborn has led to its therapeutic control. Currently, Rh-negative mothers are given anti-D globulin soon after the delivery of an Rh-positive baby. The anti-D antibodies mask the antigenic sites on the fetal red cells that may have leaked into the maternal circulation during childbirth, thus preventing long-lasting sensitization to Rh antigens.

As a result of the remarkable success achieved in prevention of Rh hemolysis, fetomaternal ABO incompatibility currently is the most common cause of immune hemolytic disease of the newborn. Although ABO incompatibility occurs in approximately 20% to 25% of pregnancies, hemolysis develops in only a small fraction of infants born subsequently, and in general the disease is much milder than Rh incompatibility. ABO hemolytic disease occurs almost exclusively in infants of blood group A or B who are born to mothers of blood group O. The normal anti-A and anti-B isohemagglutinins in group O mothers usually are of the IgM type and therefore do not cross the placenta. However, for reasons not well understood, certain group O women possess IgG antibodies directed against group A or B antigens (or both) even without previous sensitization. Therefore, the firstborn may be affected. Fortunately, even with transplacentally acquired antibodies, lysis of the infant’s red cells is minimal. There is no effective method of preventing hemolytic disease resulting from ABO incompatibility.

Nonimmune Hydrops

The major causes of nonimmune hydrops include those disorders associated with cardiovascular defects, chromosomal anomalies, and fetal anemia. Both structural cardiovascular defects and functional abnormalities (i.e., arrhythmias) may result in intrauterine cardiac failure and hydrops. Among the chromosomal anomalies, 45,X karyotype (Turner syndrome) and trisomies 21 and 18 are associated with fetal hydrops; the basis for this disorder usually is the presence of underlying structural cardiac anomalies, although in Turner syndrome there may be an abnormality of lymphatic drainage from the neck leading to postnuchal fluid accumulation (resulting in cystic hygromas). Fetal anemias due to causes other than Rh or ABO incompatibility also may result in hydrops. In fact, in some parts of the world (e.g., Southeast Asia), severe fetal anemia caused by homozygous α-thalassemia probably is the most common cause of fetal hydrops. Transplacental infection by parvovirus B19 is increasingly recognized as an important cause of fetal hydrops. The virus gains entry into erythroid precursors (normoblasts), where it replicates. The ensuing cellular injury leads to the death of the normoblasts and aplastic anemia. Parvoviral intranuclear inclusions can be seen within circulating and marrow erythroid precursors (Fig. 6–28). The basis for fetal hydrops in fetal anemia of both immune and nonimmune cause is tissue ischemia with secondary myocardial dysfunction and circulatory failure. Additionally, secondary liver failure may occur, with loss of synthetic function contributing to hypoalbuminemia, reduced plasma osmotic pressure, and edema.

Figure 6–28 Bone marrow from an infant infected with parvovirus B19. The arrows point to two erythroid precursors with large homogeneous intranuclear inclusions and a surrounding peripheral rim of residual chromatin.

  Morphology

The anatomic findings in fetuses with intrauterine fluid accumulation vary with both the severity of the disease and the underlying etiologic disorder. As previously noted, hydrops fetalis represents the most severe and generalized manifestation (Fig. 6–27), and lesser degrees of edema such as isolated pleural, peritoneal, or postnuchal fluid collections can occur. Accordingly, infants may be stillborn, die within the first few days, or recover completely. The presence of dysmorphic features suggests underlying constitutional chromosomal abnormalities; postmortem examination may reveal a cardiac anomaly. In hydrops associated with fetal anemia, both fetus and placenta are characteristically pale; in most cases, the liver and spleen are enlarged as a consequence of cardiac failure and congestion. Additionally, the bone marrow shows compensatory hyperplasia of erythroid precursors (parvovirus-associated aplastic anemia being a notable exception), and extramedullary hematopoiesis is present in the liver, the spleen, and possibly other tissues such as the kidneys, the lungs, the lymph nodes, and even the heart. The increased hematopoietic activity accounts for the presence in the peripheral circulation of large numbers of normoblasts, and even more immature erythroblasts (erythroblastosis fetalis) (Fig. 6–29).

Figure 6–29 Numerous islands of extramedullary hematopoiesis (small blue cells) are scattered among mature hepatocytes in this histologic preparation from an infant with nonimmune hydrops fetalis.

The presence of hemolysis in Rh or ABO incompatibility is associated with the added complication of increased circulating bilirubin from the red cell breakdown. The CNS may be damaged when hyperbilirubinemia is marked (usually above 20 mg/dL in full-term infants, but often less in premature infants). The circulating unconjugated bilirubin is taken up by the brain tissue, on which it apparently exerts a toxic effect. The basal ganglia and brain stem are particularly prone to deposition of bilirubin pigment, which imparts a characteristic yellow hue to the parenchyma (kernicterus) (Fig. 6–30).

Figure 6–30 Kernicterus. Severe hyperbilirubinemia in the neonatal period—for example, secondary to immune hydrolysis—results in deposition of bilirubin pigment (arrows) in the brain parenchyma. This occurs because the blood–brain barrier is less well developed in the neonatal period than it is in adulthood. Infants who survive develop long-term neurologic sequelae.

Benign Tumors

Virtually any tumor may be encountered in the pediatric age group, but three—hemangiomas, lymphangiomas, and sacrococcygeal teratomas—deserve special mention here because they occur commonly in childhood.

Hemangiomas are the most common tumors of infancy. Both cavernous and capillary hemangiomas may be encountered (Chapter 9), although the latter often are more cellular than in adults and thus may be deceptively worrisome-appearing. In children, most hemangiomas are located in the skin, particularly on the face and scalp, where they produce flat to elevated, irregular, red-blue masses; the flat, larger lesions are referred to as port wine stains. Hemangiomas may enlarge as the child gets older, but in many instances they spontaneously regress (Fig. 6–31). The vast majority of superficial hemangiomas have no more than a cosmetic significance; rarely, they may be the manifestation of a hereditary disorder associated with disease within internal organs, such as the von Hippel-Lindau and Sturge-Weber syndromes (Chapter 9). A subset of CNS cavernous hemangiomas can occur in the familial setting; affected families harbor mutations in one of three cerebral cavernous malformation (CCM) genes.

Figure 6–31 Congenital capillary hemangioma at birth (A) and at 2 years of age (B) after the lesion had undergone spontaneous regression.

(Courtesy of Dr. Eduardo Yunis, Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania.)

Lymphangiomas represent the lymphatic counterpart of hemangiomas. Microscopic examination shows cystic and cavernous spaces lined by endothelial cells and surrounded by lymphoid aggregates; the spaces usually contain pale fluid. They may occur on the skin but, more important, also are encountered in the deeper regions of the neck, axilla, mediastinum, and retroperitoneum. Although histologically benign, they tend to increase in size after birth and may encroach on mediastinal structures or nerve trunks in axilla.

Sacrococcygeal teratomas are the most common germ cell tumors of childhood, accounting for 40% or more of cases (Fig. 6–32). In view of the overlap in the mechanisms underlying teratogenesis and oncogenesis, it is interesting that approximately 10% of sacrococcygeal teratomas are associated with congenital anomalies, primarily defects of the hindgut and cloacal region and other midline defects (e.g., meningocele, spina bifida) not believed to result from local effects of the tumor. Approximately 75% of these tumors are histologically mature with a benign course, and about 12% are unmistakably malignant and lethal (Chapter 17). The remainder are designated immature teratomas, and their malignant potential correlates with the amount of immature tissue elements present. Most of the benign teratomas are encountered in younger infants (4 months of age or younger), whereas children with malignant lesions tend to be somewhat older.

Figure 6–32 Sacrococcygeal teratoma. Note the size of the lesion compared with that of the infant.

Malignant Tumors

The organ systems involved most commonly by malignant neoplasms in infancy and childhood are the hematopoietic system, neural tissue, and soft tissues (Table 6–10). This distribution is in sharp contrast with that in adults, in whom tumors of the lung, heart, prostate, and colon are the most common forms. Malignant tumors of infancy and childhood differ biologically and histologically from those in adults. The main differences are as follows:

Table 6–10 Common Malignant Neoplasms of Infancy and Childhood

CNS, central nervous system.

Many malignant pediatric neoplasms are histologically unique. In general, they tend to have a primitive (embryonal) rather than pleomorphic-anaplastic microscopic appearance, and frequently they exhibit features of organogenesis specific to the site of tumor origin. Because of their primitive histologic appearance, many childhood tumors have been collectively referred to as small, round, blue cell tumors. These are characterized by sheets of cells with small, round nuclei. The tumors in this category include neuroblastoma, lymphoma, rhabdomyosarcoma, Ewing sarcoma (peripheral neuroectodermal tumor), and some cases of Wilms tumor. Sufficient distinctive features usually are present to permit definitive diagnosis on the basis of histologic examination alone, but when necessary, clinical and radiographic findings, combined with ancillary studies (e.g., chromosome analysis, immunoperoxidase stains, electron microscopy) are used. Three common tumors—neuroblastoma, retinoblastoma, and Wilms tumor—are described here to highlight the differences between pediatric tumors and those in adults.

Neuroblastoma

The term neuroblastic includes tumors of the sympathetic ganglia and adrenal medulla that are derived from primordial neural crest cells populating these sites; neuroblastoma is the most important member of this family. It is the second most common solid malignancy of childhood after brain tumors, accounting for 7% to 10% of all pediatric neoplasms, and as many as 50% of malignancies diagnosed in infancy. Neuroblastomas demonstrate several unique features in their natural history, including spontaneous regression and spontaneous or therapy-induced maturation. Most occur sporadically, but 1% to 2% are familial, with autosomal dominant transmission, and in such cases the neoplasms may involve both of the adrenals or multiple primary autonomic sites. Germ line mutations in the anaplastic lymphoma kinase gene (ALK) recently have been identified as a major cause of familial predisposition to neuroblastoma. Somatic gain-of-function ALK mutations also are observed in a subset of sporadic neuroblastomas. It is envisioned that tumors harboring ALK mutations in either the germline or somatic setting will be amenable to treatment using drugs that target the activity of this kinase.

Morphology

In childhood, about 40% of neuroblastomas arise in the adrenal medulla. The remainder occur anywhere along the sympathetic chain, with the most common locations being the paravertebral region of the abdomen (25%) and posterior mediastinum (15%). Macroscopically, neuroblastomas range in size from minute nodules (the in situ lesions) to large masses weighing more than 1 kg. In situ neuroblastomas are reported to be 40 times more frequent than overt tumors. The great preponderance of these silent lesions spontaneously regress, leaving only a focus of fibrosis or calcification in the adult. Some neuroblastomas are sharply demarcated with a fibrous pseudocapsule, but others are far more infiltrative and invade surrounding structures, including the kidneys, renal vein, and vena cava, and envelop the aorta. On transection, they are composed of soft, gray-tan, brainlike tissue. Larger tumors have areas of necrosis, cystic softening, and hemorrhage.

Histologically, classic neuroblastomas are composed of small, primitive-appearing cells with dark nuclei, scant cytoplasm, and poorly defined cell borders growing in solid sheets (Fig. 6–33, A). Mitotic activity, nuclear breakdown (“karyorrhexis”), and pleomorphism may be prominent. The background often demonstrates a faintly eosinophilic fibrillary material (neuropil) that corresponds to neuritic processes of the primitive neuroblasts. Typically, so-called Homer-Wright pseudo-rosettes can be found in which the tumor cells are concentrically arranged about a central space filled with neuropil (the absence of an actual central lumen garners the designation “pseudo-”). Other helpful features include immunochemical detection of neuron-specific enolase and demonstration on ultrastructural studies of small, membrane-bound, cytoplasmic catecholamine-containing secretory granules.

Figure 6–33 A, Neuroblastoma. This tumor is composed of small cells embedded in a finely fibrillar matrix (neuropil). A Homer-Wright pseudo-rosette (tumor cells arranged concentrically around a central core of neuropil) is seen in the upper right corner. B, Ganglioneuromas, arising from spontaneous or therapy-induced maturation of neuroblastomas, are characterized by clusters of large cells with vesicular nuclei and abundant eosinophilic cytoplasm (arrow), representing neoplastic ganglion cells. Spindle-shaped Schwann cells are present in the background stroma.

Some neoplasms show signs of maturation, either spontaneous or therapy-induced. Larger cells having more abundant cytoplasm with large vesicular nuclei and a prominent nucleolus, representing ganglion cells in various stages of maturation, may be found in tumors admixed with primitive neuroblasts (ganglioneuroblastoma). Even better-differentiated lesions contain many more large cells resembling mature ganglion cells in the absence of residual neuroblasts; such neoplasms merit the designation ganglioneuroma (Fig. 6–33, B). Maturation of neuroblasts into ganglion cells usually is accompanied by the appearance of Schwann cells. In fact, the presence of a “schwannian stroma” composed of organized fascicles of neuritic processes, mature Schwann cells, and fibroblasts is a histologic prerequisite for the designation of ganglioneuroblastoma and ganglioneuroma; ganglion cells in and of themselves do not fulfill the criteria for maturation.

Clinical Course and Prognosis

Many factors influence prognosis, but the most important are the stage of the tumor and the age of the patient.

• Staging of neuroblastomas (Table 6–11) assumes great importance in establishing a prognosis. Special note should be taken of stage 4S (S means special), because the outlook for these patients is excellent, despite the spread of disease. As noted in Table 6–11, the primary tumor would be classified as stage 1 or 2 but for the presence of metastases, which are limited to liver, skin, and bone marrow, without bone involvement. Infants with 4S tumors have an excellent prognosis with minimal therapy, and it is not uncommon for the primary or metastatic tumors to undergo spontaneous regression. The biologic basis for this welcome behavior is not clear.

• Age is the other important determinant of outcome, and the outlook for children younger than 18 months is much more favorable than for older children at a comparable stage of disease. Most neoplasms diagnosed in children during the first 18 months of life are stage 1 or 2, or stage 4S (“low” risk category in Table 6–11), while neoplasms in older children fall into the “intermediate” or “high” risk category.

• Morphology is an independent prognostic variable in neuroblastic tumors; evidence of schwannian stroma and gangliocytic differentiation is indicative of a favorable histologic pattern.

• Amplification of the NMYC oncogene in neuroblastomas is a molecular event that has profound impact on prognosis. NMYC amplification is present in about 25% to 30% of primary tumors, most in advanced-stage disease; the greater the number of copies, the worse the prognosis. NMYC amplification is currently the most important genetic abnormality used in risk stratification of neuroblastic tumors and automatically renders a tumor as “high” risk, irrespective of stage or age.

• Deletion of the distal short arm of chromosome 1, gain of the distal long arm of chromosome 17, and overexpression of telomerase all are adverse prognostic factors, while expression of TrkA, a high-affinity receptor for nerve growth factor that is indicative of differentiation toward sympathetic ganglia lineage, is associated with favorable prognosis.

Table 6–11 Staging of Neuroblastomas

Stage 1	Localized tumor completely excised, with or without microscopic residual disease; representative ipsilateral nonadherent lymph nodes negative for tumor (nodes adherent to the primary tumor may be positive for tumor)
Stage 2A	Localized tumor resected incompletely grossly; representative ipsilateral nonadherent lymph nodes negative for tumor microscopically
Stage 2B	Localized tumor with or without complete gross excision, ipsilateral nonadherent lymph nodes positive for tumor; enlarged contralateral lymph nodes, which are negative for tumor microscopically
Stage 3	Unresectable unilateral tumor infiltrating across the midline with or without regional lymph node involvement; or localized unilateral tumor with contralateral regional lymph node involvement
Stage 4	Any primary tumor with dissemination to distant lymph nodes, bone, bone marrow, liver, skin, and/or other organs (except as defined for stage 4S)
Stage 4S*	Localized primary tumor (as defined for stage 1, 2A, or 2B) with dissemination limited to skin, liver, and/or bone marrow (<10% of nucleated cells are constituted by neoplastic cells; >10% involvement of bone marrow is considered as stage 4); stage 4S limited to infants younger than 1 year of age

* S, special.

Adapted from Brodeur GM, Pritchard J, Berthold F, et al: Revisions of the international neuroblastoma diagnosis, staging, and response to treatment. J Clin Oncol 11:1466, 1993.

Children younger than 2 years with neuroblastomas generally present with protuberant abdomen resulting from an abdominal mass, fever, and weight loss. In older children the neuroblastomas may remain unnoticed until metastases cause hepatomegaly, ascites, and bone pain. Neuroblastomas may metastasize widely through the hematogenous and lymphatic systems, particularly to liver, lungs, bones, and the bone marrow. In neonates, disseminated neuroblastomas may manifest with multiple cutaneous metastases associated with deep blue discoloration to the skin (earning the rather unfortunate moniker of “blueberry muffin baby”). As already noted, many variables can influence the prognosis of neuroblastomas, but as a rule of thumb, stage and age are the paramount determinants. Tumors of all stages diagnosed in the first 18 months of life, as well as low-stage tumors in older children, have a favorable prognosis, while high-stage tumors diagnosed in children younger than 18 months of age have the poorest outcome. About 90% of neuroblastomas, regardless of location, produce catecholamines (similar to the catecholamines associated with pheochromocytomas), which constitutes an important diagnostic feature (i.e., elevated blood levels of catecholamines and elevated urine levels of catecholamine metabolites such as vanillylmandelic acid [VMA] and homovanillic acid [HVA]). Despite the elaboration of catecholamines, hypertension is much less frequent with these neoplasms than with pheochromocytomas (Chapter 19).

Summary

Neuroblastoma

• Neuroblastomas and related tumors arise from neural crest–derived cells in the sympathetic ganglia and adrenal medulla.

• Neuroblastomas are undifferentiated, whereas ganglioneuroblastomas and ganglioneuromas demonstrate evidence of differentiation (schwannian stroma and ganglion cells). Homer-Wright pseudo-rosettes are characteristic of neuroblastomas.

• Age, stage, and NMYC amplification are the most important prognostic features; children younger than 18 months usually have a better prognosis than older children, while children with higher-stage tumors or NMYC amplification fare worse.

• Neuroblastomas secrete catecholamines, whose metabolites (VMA/HVA) can be used for screening patients.

Retinoblastoma

Retinoblastoma is the most common primary intraocular malignancy of children. The molecular genetics of retinoblastoma has been discussed previously (Chapter 5). Approximately 40% of the tumors are associated with a germline mutation in the RB1 gene and are therefore heritable. The remaining 60% of the tumors develop sporadically, and these have somatic RB1 gene mutations. Although the name retinoblastoma might suggest origin from a primitive retinal cell that is capable of differentiation into both glial and neuronal cells, it is now clear that the cell of origin of retinoblastoma is neuronal. As noted earlier, in approximately 40% of cases, retinoblastoma occurs in persons who inherit a germ line mutation of one RB allele. Familial cases typically are associated with development of multiple tumors that are bilateral, although they may be unifocal and unilateral. All of the sporadic, nonheritable tumors are unilateral and unifocal. Patients with familial retinoblastoma also are at increased risk for the development of osteosarcoma and other soft tissue tumors.

Morphology

Retinoblastomas tend to be nodular masses, usually in the posterior retina, often with satellite seedings (Fig. 6–34, A). On light microscopic examination, undifferentiated areas of these tumors are found to be composed of small, round cells with large hyperchromatic nuclei and scant cytoplasm, resembling undifferentiated retinoblasts.

Figure 6–34 Retinoblastoma.

A, Poorly cohesive tumor in retina is seen abutting the optic nerve. B, Higher-power view showing Flexner-Wintersteiner rosettes (arrow) and numerous mitotic figures.

Differentiated structures are found within many retinoblastomas, the most characteristic of these being Flexner-Wintersteiner rosettes (Fig. 6–34, B). These structures consist of clusters of cuboidal or short columnar cells arranged around a central lumen (in contrast with the pseudo-rosettes of neuroblastoma, which lack a central lumen). The nuclei are displaced away from the lumen, which by light microscopy appears to have a limiting membrane resembling the external limiting membrane of the retina.

Tumor cells may disseminate beyond the eye through the optic nerve or subarachnoid space. The most common sites of distant metastases are the CNS, skull, distal bones, and lymph nodes.

Clinical Features

The median age at presentation is 2 years, although the tumor may be present at birth. The presenting findings include poor vision, strabismus, a whitish hue to the pupil (“cat’s eye reflex”), and pain and tenderness in the eye. Untreated, the tumors usually are fatal, but after early treatment with enucleation, chemotherapy, and radiotherapy, survival is the rule. As noted earlier, some tumors spontaneously regress, and patients with familial retinoblastoma are at increased risk for the development of osteosarcoma and other soft tissue tumors.

Wilms Tumor

Wilms tumor, or nephroblastoma, is the most common primary tumor of the kidney in children. Most cases occur in children between 2 and 5 years of age. This tumor illustrates several important concepts of childhood tumors: the relationship between congenital malformation and increased risk of tumors, the histologic similarity between tumor and developing organ, and finally, the remarkable success in the treatment of childhood tumors. Each of these concepts is evident in the following discussion.

Three groups of congenital malformations are associated with an increased risk for development of Wilms tumor. Of patients with the WAGR syndrome (i.e., Wilms tumor, aniridia, genital abnormalities, and mental retardation), approximately one in three will go on to develop this tumor. Another group of patients, those with the so-called Denys-Drash syndrome (DDS), also have an extremely high risk (approximately 90%) for the development of Wilms tumor. This syndrome is characterized by gonadal dysgenesis and renal abnormalities. Both of these conditions are associated with abnormalities of the Wilms tumor 1 gene (WT1), located on 11p13. The nature of genetic aberration differs, however: Patients with WAGR syndrome demonstrate loss of genetic material (i.e., deletions) of WT1, and persons with DDS harbor a dominant negative inactivating mutation in a critical region of the gene. (A dominant negative mutation interferes with the function of the remaining wild-type allele.) The WT1 gene is critical to normal renal and gonadal development; it is not surprising, therefore, that constitutional inactivation of one copy of this gene results in genitourinary abnormalities in humans. A third group of patients, those with the Beckwith-Wiedemann syndrome (BWS), also are at increased risk for the development of Wilms tumor. These patients exhibit enlargement of individual body organs (e.g., tongue, kidneys, or liver) or entire body segments (hemihypertrophy); enlargement of adrenal cortical cells (adrenal cytomegaly) is a characteristic microscopic feature. BWS is an example of a disorder of genomic imprinting (see earlier). The genetic locus that is involved in these patients is in band p15.5 of chromosome 11 distal to the WT1 locus. Although this locus is called “WT2” for the second Wilms tumor locus, the gene involved has not been identified. This region contains at least 10 genes that normally are expressed from only one of the two parental alleles, with transcriptional silencing of the other parental homologue by methylation of the promoter region, located upstream of the transcription start site. Of all candidate “WT2” genes, imprinting abnormalities of insulin-like growth factor-2 (IGF2) have the strongest relationship to tumor predisposition in persons with BWS. IGF2 normally is expressed solely from the paternal allele, while the maternal allele is imprinted (i.e., silenced) by methylation. In some Wilms tumors, loss of imprinting (i.e., reexpression of IGF2 by the maternal allele) can be demonstrated, leading to overexpression of the IGF2 protein, which is postulated to result in both organ enlargement and tumorigenesis. Thus, these associations suggest that in some cases, congenital malformations and tumors represent related manifestations of genetic lesions affecting a single gene or closely linked genes. In addition to Wilms tumors, patients with BWS also are at increased risk for the development of hepatoblastoma, adrenocortical tumors, rhabdomyosarcomas, and pancreatic tumors. Recently, gain-of-function mutations of the gene encoding β-catenin (Chapter 5) have been demonstrated in approximately 10% of sporadic Wilms tumors.

Morphology

As seen on gross examination, Wilms tumor typically is a large, solitary, well-circumscribed mass, although 10% are either bilateral or multicentric at the time of diagnosis. On cut section, the tumor is soft, homogeneous, and tan to gray, with occasional foci of hemorrhage, cystic degeneration, and necrosis (Fig. 6–35).

Figure 6–35 Wilms tumor in the lower pole of the kidney with the characteristic tan to gray color and well-circumscribed margins.

On microscopic examination, Wilms tumors are characterized by recognizable attempts to recapitulate different stages of nephrogenesis. The classic triphasic combination of blastemal, stromal, and epithelial cell types is observed in most lesions, although the percentage of each component is variable (Fig. 6–36, A). Sheets of small blue cells, with few distinctive features, characterize the blastemal component. Epithelial “differentiation” usually takes the form of abortive tubules or glomeruli. Stromal cells are usually fibrocytic or myxoid in nature, although skeletal muscle “differentiation” is not uncommon. Rarely, other heterologous elements are identified, including squamous or mucinous epithelium, smooth muscle, adipose tissue, cartilage, and osteoid and neurogenic tissue. Approximately 5% of tumors contain foci of anaplasia (cells with large, hyperchromatic, pleomorphic nuclei and abnormal mitoses) (Fig. 6–36, B). The presence of anaplasia correlates with the presence of acquired TP53 mutations, and the emergence of resistance to chemotherapy. The pattern of distribution of anaplastic cells within the primary tumor (focal versus diffuse) has important implications for prognosis (see further on).

Figure 6–36 A, Wilms tumor with tightly packed blue cells consistent with the blastemal component and interspersed primitive tubules, representing the epithelial component. Although multiple mitotic figures are seen, none are atypical in this field. B, Focal anaplasia was present in other areas within this Wilms tumor, characterized by cells with hyperchromatic, pleomorphic nuclei and abnormal mitoses.

Nephrogenic rests are putative precursor lesions of Wilms tumors and are sometimes present in the renal parenchyma adjacent to the tumor. Nephrogenic rests have a spectrum of histologic appearances, from expansile masses that resemble Wilms tumors (hyperplastic rests) to sclerotic rests consisting predominantly of fibrous tissue with occasional admixed immature tubules or glomeruli. It is important to document the presence of nephrogenic rests in the resected specimen, since these patients are at an increased risk for the development of Wilms tumors in the contralateral kidney.

Clinical Course

Patients’ complaints usually are referable to the tumor’s enormous size. Commonly, there is a readily palpable abdominal mass, which may extend across the midline and down into the pelvis. Less often, the patient presents with fever and abdominal pain, with hematuria or, occasionally, with intestinal obstruction as a result of pressure from the tumor. The prognosis for Wilms tumor generally is very good, and excellent results are obtained with a combination of nephrectomy and chemotherapy. Anaplasia is a harbinger of adverse prognosis, but careful analyses by the National Wilms’ Tumor Study group in the United States have shown that as long as the anaplasia is focal and confined within the resected nephrectomy specimen, the outcome is no different from that for tumors without evidence of anaplasia. By contrast, Wilms tumors with diffuse anaplasia, especially those exhibiting extrarenal spread, have the least favorable outcome, underscoring the need for correctly identifying this histologic pattern.

Summary

Wilms Tumor

• Wilms tumor is the most common renal neoplasm of childhood.

• Patients with three syndromes are at increased risk for Wilms tumors: Denys-Drash, Beckwith-Wiedemann, and Wilms tumor, aniridia, genital abnormalities, and mental retardation syndrome.

• Wilms tumor, aniridia, genital abnormalities, and mental retardation syndrome and Denys-Drash syndrome are associated with WT1 inactivation, while Beckwith-Wiedemann arises through imprinting abnormalities at the WT2 locus, principally involving the IGF2 gene.

• The morphologic components of Wilms tumor include blastema (small, round blue cells) and epithelial and stromal elements.

• Nephrogenic rests are precursor lesions of Wilms tumors.

Molecular Diagnosis of Copy Number Abnormalities

As already discussed in this chapter, various diseases may occur as a result of copy number abnormalities, at the level of either the entire chromosome (trisomy 21), chromosomal segments (22q11 deletion syndrome), or submicroscopic intragenic deletions (WAGR syndrome). Karyotype analysis of chromosomes by G banding remains the classic approach for identifying changes at the chromosomal level; however, as might be expected, the resolution with this technique is fairly low. In order to identify subchromosomal alterations, both focused analysis of chromosomal regions by FISH and global genomic approaches such as comparative genomic hybridization (CGH) have become popular.

Fluorescence in Situ Hybridization (FISH)

FISH utilizes DNA probes that recognize sequences specific to chromosomal regions of greater than 100 kilobases in size, which defines the limit of resolution with this technique for identifying chromosomal changes. Such probes are labeled with fluorescent dyes and applied to metaphase spreads or interphase nuclei. The probe hybridizes to its complementary sequence on the chromosome and thus labels the specific chromosomal region that can then be visualized under a fluorescence microscope. The ability of FISH to circumvent the need for dividing cells is invaluable when a rapid diagnosis is warranted (e.g., in a critically ill infant suspected of having an underlying genetic disorder). Such analysis can be performed on prenatal samples (e.g., cells obtained by amniocentesis, chorionic villus biopsy, or umbilical cord blood), peripheral blood lymphocytes, and even archival tissue sections. FISH has been used for detection of numeric abnormalities of chromosomes (aneuploidy) (Fig. 6–37, A); for the demonstration of subtle microdeletions (Fig. 6–37, B) or complex translocations not detectable by routine karyotyping; for analysis of gene amplification (e.g., NMYC amplification in neuroblastomas); and for mapping newly isolated genes of interest to their chromosomal loci.

Figure 6–37 Fluorescence in situ hybridization (FISH). A, Interphase nucleus from a male patient with suspected trisomy 18. Three different fluorescent probes have been used in a “FISH cocktail”; the green probe hybridizes to the X chromosome centromere (one copy), the red probe to the Y chromosome centromere (one copy), and the aqua probe to the chromosome 18 centromere (three copies). B, A metaphase spread in which two fluorescent probes have been used, one hybridizing to chromosome region 22q13 (green) and the other hybridizing to chromosome region 22q11.2 (red). There are two 22q13 signals. One of the two chromosomes does not stain with the probe for 22q11.2, indicating a microdeletion in this region. This abnormality gives rise to the 22q11.2 deletion syndrome (DiGeorge syndrome).

(Courtesy of Dr. Nancy R. Schneider and Jeff Doolittle, Cytogenetics Laboratory, University of Texas Southwestern Medical Center, Dallas, Texas.)

Array-Based Genomic Hybridization

It is obvious from the preceding discussion that FISH requires previous knowledge of the one or few specific chromosomal regions suspected of being altered in the test sample. However, chromosomal abnormalities may also be detected without previous knowledge of what these aberrations may be, using a global strategy known as array-based CGH. Here the test DNA and a reference (normal) DNA are labeled with two different fluorescent dyes (most commonly, Cy5 and Cy3, which fluoresce red and green, respectively). The differentially labeled samples are then hybridized to an array of segments of genomic DNA “spotted” on a solid matrix, usually a glass slide (Fig. 6–38, A). These segments of DNA are representations of the human genome at regularly spaced intervals, and cover all 22 autosomes and the sex chromosome (Fig. 6–38, A). Amplifications and deletions in the test sample produce an increase or decrease in signal relative to the normal DNA that can be detected down to a 10-kilobase (kb) resolution (Fig. 6–38, B). Newer generations of microarrays using single-nucleotide polymorphisms (SNPs) (see further on) provide even higher resolution (with more than 1 million SNPs from the human genome on a single microarray!) and are currently being used to uncover copy number abnormalities in a variety of diseases, from cancer to autism.

Figure 6–38 Array comparative genomic hybridization (CGH) is performed by hybridization of fluorescently labeled test DNA and control DNA on a slide that contains thousands of probes corresponding to defined chromosomal regions across the human genome. The resolution with most currently available array CGH technology is on the order of approximately 10 kb. A, Higher-power view of the array demonstrates copy number aberrations in the test Cy5-labeled sample (red), including regions of amplification (spots with excess of red signal) and deletion (spots with excess of green signal); yellow spots correspond to regions of normal (diploid) copy number. B, The hybridization signals are digitized, resulting in a virtual karyotype of the genome of the “test” sample. In the illustrated example, array CGH of a cancer cell line identifies an amplification on the distal long arm of chromosome 8, which corresponds to increased copy number of the oncogene MYC.

(A, From Snijders AM, Nowak N, Segraves R, et al: Assembly of microarrays for genome-wide measurement of DNA copy number. Nat Genet 29:263, 2001; Web Figure A. Copyright 2001. Reprinted by permission from Macmillan Publishers Ltd.)

Direct Detection of DNA Mutations by Polymerase Chain Reaction (PCR) Analysis

PCR analysis, which involves exponential amplification of DNA, is now widely used in molecular diagnosis. If RNA is used as the substrate, it is first reverse-transcribed to obtain cDNA and then amplified by PCR. This method involving reverse transcription (RT) often is abbreviated as RT-PCR. One prerequisite for direct detection is that the sequence of the normal gene must be known. To detect the mutant gene, two primers that bind to the 3′ and 5′ ends of the normal sequence are designed. By utilizing appropriate DNA polymerases and thermal cycling, the target DNA is greatly amplified, producing millions of copies of the DNA sequence between the two primer sites. The subsequent identification of an abnormal sequence can then be performed in several ways:

Figure 6–39 Microarray-based DNA sequencing. Left panel, A low-power digitized scan of a “gene chip” that is no larger than a nickel in size but is capable of sequencing thousands of base pairs of DNA. High-throughput microarrays have been used for sequencing whole organisms (such as viruses), organelles (such as the mitochondria), and entire human chromosomes. Right panel, A high-resolution view of the gene chip illustrates hybridization patterns corresponding to a stretch of DNA sequence. Typically, a computerized algorithm is available that can convert the individual hybridization patterns across the entire chip into actual sequence data within a matter of minutes (“conventional” sequencing technologies would require days to weeks for such analysis). Here, the sequence on top is the reference (wild-type) sequence, while the lower one corresponds to the test sample sequence. As shown, the computerized algorithm has identified a C→G mutation in the test sample.

(Adapted from Maitra A, Cohen Y, Gillespie SE, et al: The Human MitoChip: a high-throughput sequencing microarray for mitochondrial mutation detection. Genome Res 14:812, 2004.)

Figure 6–40 Allele-specific polymerase chain reaction (PCR) analysis for mutation detection in a heterogeneous sample containing an admixture of normal and mutant DNA. Nucleotides complementary to the mutant and wild-type nucleotides at the queried base position are labeled with different fluorophores such that incorporation into the resulting PCR product yields fluorescent signals of variable intensity in accordance with the ratio of mutant to wild-type DNA present.

Figure 6–41 Principle of next-generation sequencing. Several alternative approaches currently are available for “NextGen” sequencing, and one of the more commonly used platforms is illustrated. A, Short fragments of genomic DNA (“template”) between 100 and 500 base pairs in length are immobilized on a solid phase platform such as a glass slide, using universal capture primers that are complementary to adapters that have previously been added to ends of the template fragments. The addition of fluorescently labeled complementary nucleotides, one per template DNA per cycle, occurs in a “massively parallel” fashion, at millions of templates immobilized on the solid phase at the same time. A four-color imaging camera captures the fluorescence emanating from each template location (corresponding to the specific incorporated nucleotide), following which the fluorescent dye is cleaved and washed away, and the entire cycle is repeated. B, Powerful computational programs can decipher the images to generate sequences complementary to the template DNA at the end of one “run,” and these sequences are then mapped back to the reference genomic sequence, in order to identify alterations.

(Reproduced with permission from Metzker M: Sequencing technologies—the next generation. Nat Rev Genet 11:31–46, 2010, © Nature Publishing Group.)

Linkage Analysis and Genome-Wide Association Studies

Direct diagnosis of mutations is possible only if the gene responsible for a genetic disorder is known and its sequence has been identified. In several diseases that have a genetic basis, including some common disorders, direct genetic diagnosis is not possible, either because the causal gene has not been identified or because the disease is multifactorial (polygenic) and no single gene is involved. In such cases, two types of analyses can be performed for unbiased identification of disease-associated gene(s): linkage analysis and genome-wide association studies (GWASs). In both instances, surrogate markers in the genome, also known as marker loci, must be used to localize the chromosomal regions of interest, based on their linkage to one or more putative disease-causing genes. The marker loci utilized are naturally occurring variations in DNA sequences known as polymorphisms. The most common DNA polymorphisms—SNPs—occur at a frequency of approximately one nucleotide in every 1000–base pair stretch and are found throughout the genome (e.g., in exons and introns and in regulatory sequences). SNPs serve both as a physical landmark within the genome and as a genetic marker whose transmission can be followed from parent to child.

Two technologic breakthroughs have enabled the application of SNPs to high-throughput “gene hunting”: First is the completion of the so-called HapMap project, which has provided linkage disequilibrium patterns in three major ethnoracial groups, based on genome-wide SNP mapping. The entire human genome can now be divided into blocks known as “haplotypes,” which contain varying numbers of contiguous SNPs on the same chromosome that are in linkage disequilibrium and hence inherited together as a cluster. As a result, rather than querying every single SNP in the human genome, comparable information about shared DNA can be obtained simply by looking for shared haplotypes, using single or a small number of SNPs that “tag” or identify a specific haplotype. Second, it is now possible to simultaneously genotype hundreds of thousands to a million SNPs at one time, in a cost-effective way, using high-density SNP chip technology.

Indications for Genetic Analysis

The preceding discussion described some of the many techniques available today for the diagnosis of genetic diseases. For judicious application of these methods, it is important to recognize which persons require genetic testing. In general, genetic testing can be divided into prenatal and postnatal analysis. It may involve conventional cytogenetics, FISH, molecular diagnostics, or a combination of these techniques.

Prenatal genetic analysis should be offered to all patients who are at risk of having cytogenetically abnormal progeny. It can be performed on cells obtained by amniocentesis, on chorionic villus biopsy material, or on umbilical cord blood. Some important indications are the following:

Postnatal genetic analysis usually is performed on peripheral blood lymphocytes. Indications are as follows:

In the context of these and other clinical applications, an important point is that we are currently living in a breath-taking era of so-called genomic medicine. Future years will foretell how the advances in elucidation of the genetic basis of human disease will affect its diagnosis, prevention, and therapy.