| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Metabolic Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892-1802
Correspondence: Address all correspondence and requests for reprints to: Stephen J. Marx, M.D., Building 10, Room 9C-101, National Institutes of Health, 10 Center Drive, MSC 1802, Bethesda, Maryland 20892-1802. E-mail: StephenM{at}intra.niddk.nih.gov
 |
Abstract |
|---|
 Top Abstract I. Introduction: Frameworks for...II. Heredity Contributes to...III. The Gene for...IV. Hereditary Hormonal...V. Relations of Monoclonal...VI. Tissue Specificity in...VII. Syndrome and Gene...VIII. ConclusionReferences |
|---|
|
I. Introduction: Frameworks for Organizing Syndromes of Hormonal Excess |
|---|
|
TopAbstractI. Introduction: Frameworks for...II. Heredity Contributes to...III. The Gene for...IV. Hereditary Hormonal...V. Relations of Monoclonal...VI. Tissue Specificity in...VII. Syndrome and Gene...VIII. ConclusionReferences |
|---|
Hereditary hormone excess is usually part of a primarily multiorgan syndrome. The grouping of affected tissues in a syndrome usually has not matched an intuitive or otherwise logical pattern. Several broad classification frameworks have been proposed to help deal with this concern.
A. Older frameworks
Recognition of familial neoplasms2 was first reported in the 1800s (1) as tuberous sclerosis complex (TSC) and peripheral neurofibromatosis (NF) type 1 (NF1) (2, 3, 4). Obviously, a component of a syndrome could not have been understood as hormonal until after the major discoveries of hormones, which did not begin until the early 1900s. Heredity as a feature of hormonal excess was not well-recognized until its extended characterization in multiple endocrine neoplasia (MEN) type 1 (MEN1) in 1954 (5, 6, 7). Heredity in von Hippel-Lindau disease (VHL) was not established until a decade later (8).
Although the concepts of heredity have been progressively strengthened, certain terms to organize these disorders have not proved durable. The category of "phakomatosis" was derived from the Greek phakos for birthmark; this term was developed originally to categorize syndromes with tumors of the central nervous system and skin, such as TSC and peripheral NF (9). The clinical similarities in these syndromes suggested that they had related causes. This classification remains in occasional use today, although with many modifications (10). An authoritative, current definition of phakomatosis illustrates its acquired complexity: "conditions that predispose to hamartomas and other tumors and involve the skin and/or eyes, nervous system, and one or more body systems" (11). No common cause among the phakomatoses has been recognized.
The method of argentaffin or silver-based staining was initially developed in 1870 (12) and supported a histological classification method. Argentaffin staining, a rough correlate with neuroendocrine origin, was cited as a defining feature of several tumors, including several tumors in MEN1 and MEN2 (13). It was already evident in 1970 that tumors of "neuroectoderm" origin were not all positive for the argentaffin stain. The nonspecific nature of argentaffin methods derives from histochemical reactivity with selected small amines (14).
Pearse (15, 16) and others promoted a related classification, based on histochemical and developmental concepts, designating certain tissues as APUD, based on their potential for amine precursor uptake and decarboxylation. Subsequently, this was expanded into the concept of a "neuroendocrine system." The APUD classification has long been questioned by some embryologists as an oversimplification, which did not account for valid embryological origin of certain structures (17).
At least one related concept has persisted. Certain tumors in MEN2 (C cell neoplasia and adrenal medullary neoplasia) are attributed to an origin from one neuroendocrine structure, specifically the neural crest (18, 19); however, the inherent parathyroid tumors seem to conflict with any suggestion of unitary origin of all MEN2A tumors from neural crest (17).
Nesidioblastosis was proposed as the basis for pancreatic islet tumors (20). As with the APUD concept, this concept was based in histology and embryology. Nesidioblastosis is a normal embryonic process in which pancreatic ductules give rise to buds that could develop into islets. When pathologically increased, nesidioblastosis might be central in hypersecretion of insulin, other hormones, and other factors (21). Nesidioblastosis was considered as possibly contributing to two states that were incompletely separated: persistent hyperinsulinemic hypoglycemia of infancy (PHHI), and pancreatic islet macro- and microadenomas in adults with and without MEN1. More recent evaluations have found that nesidioblastosis is sometimes a feature of the normal pancreas at all ages. Also, it has not been confirmed as characteristic within any syndrome. Histopathological interpretation of abnormalities in islet hyperfunction syndromes has shifted to other features (22, 23, 24).
Tumor multiplicity was recognized early in many neoplasia syndromes and has also supported concepts of humoral contributions to tumorigenesis. In particular, the islets or other tissues in MEN1 have been suggested to release factors that acted via the bloodstream to promote extrapancreatic hormonal tumors (25, 26). Limited support came from suggestions of gastrin acting as a trophic factor for gastric carcinoid tumors (27) and for a humoral role of epidermal growth factor-like peptides in NF1 (28). Lastly, GH-secreting tumors promote distant release of IGF-I, which arguably exerts its own tumorigenic effects (29). The other hypothesized, humoral factors in the multiple neoplasias have not been fully identified. Subsequent evidence, however, favored a more central role for intrinsic defects, such as multifocal independent monoclonal expansions,3 as the major explanation for tumor multiplicity (30, 31, 32).
B. Current frameworks and syndrome coverage
Current frameworks for states of hereditary hormone excess are more anatomical and more etiologically neutral than several earlier ones. Terms like isolated hyperparathyroidism, neuroendocrine, enteropancreatic, and multiple neoplasia are typical. Similarly, the continued use of eponyms for a few syndromes is noncommittal about etiology. At an increasing rate, the molecular basis of many syndromes is being clarified. Important aspects have been reviewed recently (33, 34, 35, 36).
This paper covers the following topics about hereditary hormonal excess: processes of tumorigenesis, mutated genes, their encoded defective molecules, disturbed pathways, and tissue patterns of expression (particularly hormonal expressions). A major focus of this paper is the hormone secretory (see Footnote 1) aspects of certain syndromes. To include relevant exceptions, this coverage includes hereditary syndromes with gland overgrowth and an often unrealized potential for hormone oversecretion, such as thyroid cancer, carcinoid tumor, and other so-called nonfunctional hormonal tumors (islet, pituitary, adrenal cortex). Conditions mimicking hormone excess but arising from mutation in hormone target tissues are not included in this review. For example, Liddle syndrome is omitted; this can arise from activating mutation in either of two subunits of the epithelial sodium channel, mimicking but not associated with hyperaldosteronism (37). Similarly, tissue hyperfunction resulting from upstream regulation, such as ACTH action with steroidogenesis defects, is not covered.
|
II. Heredity Contributes to Gene Identification |
|---|
|
TopAbstractI. Introduction: Frameworks for...II. Heredity Contributes to...III. The Gene for...IV. Hereditary Hormonal...V. Relations of Monoclonal...VI. Tissue Specificity in...VII. Syndrome and Gene...VIII. ConclusionReferences |
|---|
). Furthermore, the familial basis of a tumor may be usually obvious as in gastrinoma or usually occult as in pheochromocytoma (40, 41).
|
A. Broad routes to identification of a gene that is tumorigenic in the germline
1. Purification of a biological activity associated with a protein or a cDNA
a. Functional cloning.
Identification of a disease gene before the middle 1980s relied mainly on functional cloning (44): gene derivation after biochemical purification of a protein associated with a special biological function—for example, a pigment (hemoglobin), a peptide hormone, or an enzyme.
b. Expression cloning.
In expression cloning, a detectable specific biological activity, such as that from the calcium-sensing receptor (cas-r), in a pool of many cDNAs, was expressed and measured in transfected reporter cells; then, the activity was purified by serial enrichment, expression, and repeated selection among cDNA pools (45).
Functional cloning or expression cloning was never possible for identification of most tumorigenic genes, because an underlying biochemical disturbance could not be predicted or recognized and then purified before identification of the gene. However, selected tumorigenic genes were identified through their broad oncogenic properties that they could transmit to susceptible reporter cells. In particular, mouse NIH-3T3 cells were exploited for their ability to be "transformed" by so-called direct acting oncogenes. These methods allowed the identification of RET, MET, RAS, RAF, SRC, SIS, MYC, FOS, JUN, and a limited number of other oncogenes (46). Many oncogenes have been implicated to varying degrees in sporadic tumors. Few (such as RET, MET, and CDK4) have been implicated in hereditary tumors.
Although some oncogenes could thus be isolated in susceptible cells, no equivalent system has been developed for identifying the many tumor suppressor genes. Their shared tumor suppressor property would inhibit cell accumulation and, thereby, impair enrichment by cell accumulation.
2. Precise subchromosomal mapping of the disease trait
a. Positional cloning.
The first, generally applicable, mapping approach was positional cloning (47). In positional cloning, the subchromosomal interval of candidate genes is narrowed as much as possible to allow the subsequent steps; these are the cloning of all or most genes in the narrowed candidate gene interval and then sequencing among these to find the defining mutations in a panel of germline DNAs from index cases.
b. Positional candidate approach.
An alternative approach has been the positional candidate approach (47). First, the subchromosomal interval of candidate genes is narrowed as above. Because some or even all genes in the candidate interval had already been characterized in part, those that seem most promising as candidates are tested selectively for mutations. Both of these approaches became more powerful as the map of the human genome approached completion. Still, only about half of unidentified disease genes can be predicted from appealing candidate genes. The remainder have uninformative sequence (48). If the positional candidate approach is not successful, positional cloning remains as the default method.
B. Heredity in a tumor syndrome supports two independent mapping techniques for narrowing the subchromosomal interval of candidate genes
1. Reconstruction of meiotic recombination events from haplotypes.
Haplotype analysis in families permits reconstruction and analysis of an archival meiotic crossover (same as meiotic recombination) event in development of a gamete. A meiotic crossing-over event on the gamete carrying the mutated candidate gene can establish, near the candidate gene locus, a boundary in the chromosomal map of that gamete. All markers on one side of the boundary segregate with the disease phenotype. All markers on the other side of the boundary are traceable to an unaffected ancestor and thus define a zone of excluded candidate genes. A candidate interval can be narrowed by retrospective analysis of haplotype maps of many gametes to reconstruct the progressively rare crossing-over, closer and closer to the candidate gene. Such methods can narrow a candidate interval to one million bases (about 1 cM).
An analogous evaluation can be done among multiple families with the same syndrome. If several families are discovered to share a statistically significant and disease-associated core haplotype including the candidate gene locus, it generally means that they inherited that haplotype with the associated disease-predisposing mutation from a common ancestor with that mutation (49). This core haplotype, then, becomes conceptually similar to boundaries from meioses from crossing-over events in a single extended kindred.
2. Tumoral loss of heterozygosity about the locus of candidate genes.
The first step in tumorigenesis is often a single base DNA change or other small mutation of one allele within a tumorigenic gene; this mutation is present in every cell if it was acquired through germline transmission but may cause no phenotype in the subject or in the cell. If the gene contributes to tumor development by a loss-of-function mechanism, a second step in tumorigenesis is often a large subchromosomal or whole chromosomal rearrangement that removes the remaining normal copy of that same gene along with the contiguous copies of many genes on either side from the same DNA strand (32). This reduction to the null status of the syndromal gene frees one tumor precursor cell to proliferate into a tumor clone; the same large DNA rearrangement is thus present in all cells of the tumor clone. Similar biallelic events in one tumor precursor cell can lead to a hereditary or a nonhereditary tumor clone.
The triggering rearrangement can be recognized from its outcome as a loss of alleles from the previously normal copy of the chromosome in the tumor DNA when a subchromosomal DNA marker is heterozygous in the germline (usually tested with leukocyte DNA as a germline surrogate); this is termed loss of heterozygosity (LOH) or allelic loss about the locus of that gene.
If the tumor syndrome is hereditary, if multiple syndromic tumors are tested, and if there is consistent LOH at the presumed germline locus, then there is a high likelihood that the LOH of every tumor with LOH at this locus gives boundaries that bracket the sought gene. This method can be strengthened by haplotype analysis involving comparison of tumor DNA with other tumor DNA or with germline DNA of several family members to establish that allelic losses were from the chromosome of the unaffected parent (49). If a tumor is not hereditary, there must be uncertainty about the identity of the gene(s) with the first hit within that locus. A substantial fraction of all genes, through a loss of function, can contribute to or coexist with cell accumulation. If the initial copy with loss of function does not involve the gene of interest, then any derived boundary for LOH near that locus can give information that would be misleading for identification of that syndromal gene.
C. Panel of germline DNAs from several index cases often holds several different mutations in one sought disease gene: sequencing that gene in the panel and finding the mutations is a powerful criterion for disease gene identification
Obviously, the final step in identification of a disease gene is proof that mutation in one candidate gene causes the disease. The usual criterion is that mutation therein associates clearly with the rare and specific syndrome. Most of the genes described in this paper were identified through a combination of subchromosomal mapping and then identification by the demonstration of disease-associated mutations. When a family with a well-characterized rare syndrome is identified, there is high likelihood that the family has mutation in the same gene. Typically, a panel of germline DNAs is compiled with one index case from each of several such kindreds and with no suspicion of common ancestry, making it less likely that the same mutation is represented more than once. Although phenocopies of a syndrome sometimes exist based on mutation in other genes (such as with at least four different genes whose mutation causes PHHI; see Section IV.E.5.c), phenocopy can occasionally be excluded in a family by demonstrating genetic linkage to the candidate locus. Not surprisingly, there is at least one relevant exception to the specificity of this application of genetic linkage: SUR1 or Kir6.2 are not homologs, but either can be mutated in indistinguishable families with PHHI, and each maps to approximately the same location in chromosome 11p15 (see Section IV.E.5.e).
|
III. The Gene for a Hereditary Syndrome Often Contributes to Sporadic Tumors |
|---|
|
TopAbstractI. Introduction: Frameworks for...II. Heredity Contributes to...III. The Gene for...IV. Hereditary Hormonal...V. Relations of Monoclonal...VI. Tissue Specificity in...VII. Syndrome and Gene...VIII. ConclusionReferences |
|---|
) (see Section VI.A). Of course, the specificity also pertains to the group of organs in which a given mutated gene can contribute to tumors. For example, the MEN1 gene contributes to either hereditary or sporadic tumors in the parathyroids, the pancreatic islets, bronchial carcinoid, and certain other tissues (31). Similarly, the RET gene contributes to hereditary tumor and to a lower fraction of sporadic tumors, in thyroidal C cells and adrenal medulla (50).
2. Degree of morbidity, particularly cancer and its aggressiveness.
Aggressiveness, as manifested by cancer potential, earlier age at tumor onset, and/or embryonal lethality, is inherent in certain mutated genes and in selected mutations of certain genes. For example, loss of function of the hSWI5/INI1 gene contributes to atypical teratoid/rhabdoid tumors of the central nervous system and several other nonhormonal tumors that are malignant in the first year of life in either sporadic or familial tumors (51, 52).
Similar trends toward aggressive expressions are seen with a specific type of mutation in some hereditary or nonhereditary hormonal neoplasms. For example, the germline methionine-918 mutations in RET are associated with highly aggressive and early-onset medullary thyroid cancer (MTC) in the MEN2B syndrome. Among sporadic MTC, the identical mutation in somatic DNA is also associated with more aggressive tumor (53, 54) (see Sections IV.A.5 and IV.C.2).
B. One mutation can have different expressions in hereditary or sporadic tumors
1. Tumor multiplicity in hereditary tumors.
Unlike sporadic tumors, many hereditary tumors, particularly those with high penetrance, express multiplicity. In fact, lack of tumor multiplicity within an organ is unusual and puzzling in a highly penetrant hereditary tumor, such as the pituitary in MEN1 (55). Multiplicity is recognized by any one among several criteria. First, the term "multiple" describes tumors in differing organs within a syndrome such as the parathyroid, pancreatic islets, and the pituitary in MEN1 (31). Second, multiplicity describes tumor in separate portions of one tissue that normally is anatomically discontinuous. This multiplicity is sometimes striking in a dispersed organ such as the pancreatic islets, the duodenal endocrine tissues, or the parathyroids or in a paired organ such as the adrenal cortex or adrenal medulla (56). Third, tumors can be multiple even within a continuous tissue. When expressed as tumor nodules, this is readily recognizable as in the thyroid (57, 58). But when not nodular, this type of multiplicity may not be detected without use of special analyses, such as microdissection (56).
2. Earlier age at onset of hereditary tumors.
For certain tumors, age of onset is earlier in the hereditary than the nonhereditary setting. The age differential may be striking, absent, or even reversed. For example, comparing tumors in MEN1 cases vs. sporadic cases, the difference in age of onset for parathyroid tumors is 30 yr (age 25 vs. 55 yr); for gastrinomas it is only 10 yr (age 35 vs. 45 yr), but for prolactinoma there is no age difference (age 35 yr in either) (31, 59, 60). This age differential can occur from tumor suppressor genes or oncogenes. For example, the approximate age of onset for palpable MTC for untreated cases with MEN2B (i.e., mainly mutated RET methionine-918 in the germline) vs. in sporadic cases (25% same mutation, somatically) is age 20 vs. 50 yr (61). This age differential is not seen for RET mutations that cause familial MTC (FMTC).
3. Differences in tissue selectivity.
Certain genes or even certain mutations contribute to hereditary or sporadic tumors of the same tissue but in widely differing proportions. For example, RET germline mutation is associated with parathyroid tumor in 30–60% of cases with MEN2A, but no similar or other somatic mutation of the RET gene has so far been identified in a sporadic parathyroid tumor (62, 63). Similarly, MEN1 biallelic inactivation is implicated in most uterine fibroids of women with MEN1 but rarely if ever in sporadic uterine fibroid tumor (64). Conversely, somatic P53 mutation in serine-249 is frequent in hepatocellular cancer, likely reflecting direct consequences of aflatoxin exposure in somatic tissue (65); in contrast, liver cancer and particularly germline mutation of this codon are rare in Li-Fraumeni syndrome.
4. Fatal to the embryo.
For certain genes, selected mutations, which contribute to sporadic tumors, are not seen in hereditary tumors, raising the possibility that, if that mutation occurred in the germline, it would be lethal during embryogenesis. This may apply to several tumorigenic mutations of some tumor suppressors and to most known mutations in oncogenes. For example, RAS mutation is found in half of colon cancers (sporadic or hereditary); however, RAS mutation has not been found to be transmitted through the germline (66, 67).
|
IV. Hereditary Hormonal Excess—Syndromes, Genes, and Molecules |
|---|
|
TopAbstractI. Introduction: Frameworks for...II. Heredity Contributes to...III. The Gene for...IV. Hereditary Hormonal...V. Relations of Monoclonal...VI. Tissue Specificity in...VII. Syndrome and Gene...VIII. ConclusionReferences |
|---|
A syndrome often is pathognomonic of germline mutation in just one or a very small number of alternate genes. This section of the paper reviews the pairing between syndromes, their mutated genes, and their mutant molecules. A later section covers relations between syndromes, their disturbed molecular pathways, and hormone excess.
A. Hereditary monoclonal excess limited to one hyperfunctioning hormonal tissue
1. Isolated hyperparathyroidism
a. Expressions.
A syndrome of familial isolated hyperparathyroidism (FIH), unrelated to incomplete expression of another syndrome, has not been fully documented in a large kindred. In theory, FIH could reflect incomplete expression of any of four or more complex syndromes as follows: MEN1, MEN2A, familial hypocalciuric or benign hypercalcemia (FHH), or hyperparathyroidism-jaw tumor syndrome (HPT-JT) (70). In surveys of FIH, the etiologies of about one third are somewhat evenly divided between incomplete expressions of MEN1, FHH, or HPT-JT (70, 71). The remaining fraction of families with FIH could have other undiscovered syndromes or "true" FIH.
Each of the two largest reported families with FIH and MEN1 mutation had 14 affected or unaffected carriers (72, 73). Based on a few large kindreds, there is a formal possibility that isolated hyperparathyroidism from mutation of MEN1 could be a durable phenotype. In analogy, another unusual and large family with CASR mutation but not expressing FHH has an FIH phenotype, with hypercalciuria, stones, monoclonal parathyroid adenomas, and benefit from subtotal parathyroidectomy (74, 75).
b. Pathology.
In FIH, there has generally been involvement of multiple parathyroid glands whether or not synchronous. Parathyroid tumors include polyclonal features, monoclonal features, cystic features in few, and even rare parathyroid malignancy (70). This heterogeneity undoubtedly reflects the unrecognized inclusion of several distinct etiologies within FIH.
c. Genes and loci.
No gene unique for nonsyndromic isolated hyperparathyroidism has been identified or even mapped. Most families have been too small for genomewide linkage analyses. Virtually each of the few large kindreds with FIH has had a recognized germline mutation in the MEN1, CASR, or HRPT2 gene (70, 72, 76).
d. Molecules.
Menin, the cas-r, and parafibromin are discussed with their full syndromes (see Sections IV.B.1, IV.C.1, IV.E.1, and IV.E.2).
2. Isolated pituitary tumor
a. Expressions.
Isolated pituitary tumor has been found in about 100 families (77, 78). Only about five such families are as large as four to nine affected plus unaffected carriers (77, 79). The pituitary tumors are generally GH-secreting. A small fraction of cases have prolactinoma. Because sporadic GH-secreting tumors can cosecrete prolactin, and because prolactinomas seem randomly distributed in the families with GH-secreting tumor, those families with variable fraction of prolactinoma are currently classified as part of this single entity.
b. Pathology.
The histopathology of familial pituitary tumor is benign and indistinguishable from that of sporadic tumor. 11q13 LOH has been seen in the majority of tested tumors (77) supporting mono- or oligoclonality.
c. Genes and loci.
A causative gene has not been identified for any family with isolated pituitary tumor. Germline sequencing of MEN1 or PRKAR1A has been done in an index case from many families without finding any mutation (77, 78, 80, 81). Genetic linkage analysis has been reported in only two large families and has been suggested in each gene at 11q13 (thus, near or even identical to the MEN1 gene) (77, 79). LOH testing in the familial tumors has also been compatible with germline or somatic mutation in a tumor suppressor gene at 11q13 (77).
3. Isolated paraganglioma or pheochromocytoma
a. Expressions.
Paraganglioma is a vascular, usually benign tumor, arising from extraadrenal tissue and associated with the parasympathetic nervous system (82, 83). Paraganglioma is distinguished from pheochromocytoma by its location and its chemical properties. Paraganglioma is typically in the head or neck about the carotid body, the glomus bodies of the jugular bulb, the tympanic plexus, or the vagus nerve. The clinical features depend on the site of origin. Most symptoms of paraganglioma relate to cranial palsies and other mass effects. Uncommonly (about 1%), symptoms arise from systemic catecholamine excess. Familial transmission is recognizable in 10% of paraganglioma cases, and cause by a germline mutation is even more common. Most of the tumors are paraganglioma, but pheochromocytoma is possible in these families with a varying frequency that depends on the mutated gene.
Pheochromocytoma occurs in the adrenal medulla and paraadrenal regions, particularly in the organs of Zuckerkandel. Familial isolated pheochromocytoma (FIPh) has at least nine potential causes: MEN2A, VHL, isolated paraganglioma/pheochromocytoma syndromes from mutation in any of four genes (PGL1–PGL3 are three similar syndromes; and SDHB), NF1, MEN1, and Carney triad. However, among these nine, only VHL type IIC is claimed to be a robust syndrome of isolated pheochromocytoma. The first six have been recognized as occasional causes of FIPh. NF1 and MEN1 should present rarely if ever as FIPh, because other features of each are much more highly penetrant. Incomplete expression of Carney triad as FIPh cannot currently be recognized in the absence of even a mapped locus for that syndrome (see Section IV.B.8). Hereditary pheochromocytoma is bilateral by adulthood in 30–50% of cases but only in those syndromes where the penetrance of pheochromocytoma is high.
b. Pathology.
The microscopic appearance of paraganglioma or pheochromocytoma is similar. Both are vascular, perhaps due in part to the vascular endothelial growth factor (VEGF) that they overexpress (84). Some 5–10% of paragangliomas or pheochromocytomas become malignant. Malignant behavior correlates poorly with histological appearance.
c. Genes and loci.
Two familial paraganglioma variants (PGL1 caused by SDHD at 11q22.3, and PGL3 caused by SDHC at 1q21) are transmitted directly from the father, with generation skipping when derived from the mother. This represents an unusual silencing/imprinting mechanism, because the same unaffected and silenced maternal allele seems to be inactivated somatically during tumorigenesis. A third variant (PGL with unassigned number and caused by SDHB at 1p36) differs insofar as it may be only transmitted maternally and has mainly pheochromocytoma. One family classified as PGL2 maps to 11q13, and its mutated gene has not been identified.
Germline mutations were searched among 271 cases with apparently sporadic pheochromocytoma. There was germline mutation in 24% overall, with germline mutation of VHL in 11%, RET in 4%, SDHD in 4%, and SDHB in 4%; SDHC was not tested (40). Somatic mutation of RET, VHL, SDHB, or SDHD is only 1–10% each in sporadic pheochromocytoma (85).
The mutation patterns in SDHD, SDHC, or SDHB predict loss of function of the encoded protein. LOH at chromosome 11 is usually evident in FIPh tumors and in some 10% of sporadic paragangliomas. This correlates with the locus of SDHD, the main contributing tumor suppressor gene; MEN1 and the gene for PGL2 are also on chromosome 11 but are not known to contribute to a large fraction of these tumors.
d. Molecules.
Succinate dehydrogenase, also known as succino-ubiquinone oxidoreductase, is the mitochondrial complex II enzyme with four subunits; a separate nuclear gene encodes each subunit. The paraganglioma syndrome has been attributed to an inactivating mutation in three of the four succinate dehydrogenase (sdh) subunits: sdhd, sdhc, or sdhb.
4. Isolated nonmedullary cancer of the thyroid
a. Expressions.
Sporadic nonmedullary cancer of the thyroid is common. Very few large families that suggest autosomal dominant transmission have been reported (39). Its occurrence in only a few members of a family is thus probably not related to a monogenic cause. Hereditary thyroid cancer at low penetrance can also be an expression from germline mutation in PTEN, APC, WRN, or PRKAR1A (Table 1), but none of these mutated genes is likely to present as isolated thyroidal cancer; each has other expressions that are far more penetrant.
b. Pathology.
Like sporadic tumors, hereditary nonmedullary thyroid tumors are mainly papillary. Principally oxyphilic thyroid tumors were noted in one of the two families with linkage to 19p13.2. The hereditary thyroid tumors with polyposis of the colon show excess of cribriform pattern, whereas those in Cowden syndrome or Werner syndrome show excess of follicular pattern (see Sections IV.B.2 and IV.B.4).
c. Genes and loci.
No mutant gene has so far been identified. Three chromosomal loci for candidate genes for isolated thyroid cancer have been identified, each from genetic linkage testing in one or two large families: NMTC at 2q21, TCO at 19p13.2 [thyroid tumors with oxyphilia (86)], and PTC at 1q21 [with lower penetrance of papillary renal neoplasia (39)]. The locus at 2q21 was delineated in a large Tasmanian family and then supported in 17 of 80 smaller families (87). Dominant multinodular goiter has been seen in two large families with mapping to 14q and Xp22 (39).
5. Isolated MTC
a. Expressions.
Familial isolated MTC (FMTC), by definition, lacks other clinical features of MEN2A or MEN2B (50). Furthermore, cases with FMTC have MTCs that begin at later age and with lower morbidity than MTC with MEN2A or MEN2B. The differences in expression from MEN2A are particularly striking when only the larger families with FMTC are considered (54, 88, 89, 90). Ten percent of cases with FMTC also expressed papillary thyroid cancer, which is high even considering the high rate of discovery incidental to surgery for MTC (91).
FMTC can be classified through its phenotype or through its germline mutated RET gene (92). In either case, it can be considered as a variant of MEN2A although it is not literally a MEN syndrome. In several reports, the criteria for FMTC are not stated, and the inclusion of small kindreds increases the likelihood of some families with incomplete MEN2A. In order not to omit the different management appropriate for MEN2A, a definition has been recommended, requiring any FMTC kindred to have over 10 affected members (93). Still the separation of FMTC from MEN2A seems imperfect. For example, one family with RET Val804Leu mutation has met the rigorous size criterion for FMTC (54), but two pheochromocytomas have been reported in another family with the same RET mutation (94).
b. Pathology.
The pathology of the stages of C cell cancer in FMTC is indistinguishable from that in MEN2A or MEN2B.
c. Genes and loci.
Isolated MTC has been reported in about 50 families. All but a few families have an identified germline RET mutation (see Section IV.C.2.c). Mutation in the following RET codons has been almost fully specific for FMTC: 630, 768, 790, 791, 804 (Fig. 1) (50). Some of these same RET codon germline mutations have been found in sporadic MTC or sporadic pheochromocytoma (50).
|
d. Molecules.
Unlike for the RET mutations in MEN2A and MEN2B (see Section IV.C.2.d), no unique molecular function or molecular domain has been assigned to these RET mutations, possibly specific for FMTC. However, the overall transforming potency in vitro has been lower with mutations associated with FMTC than with mutations causing other MEN2 variants (96).
6. Isolated adrenal-cortical tumor
a. Expressions.
Autosomal dominant isolated hyperfunction of the adrenal cortex can cause only hypercortisolism or only hyperaldosteronism. Hereditary adrenocortical tumor can also be an expression of MEN1, Carney complex, or Li-Fraumeni syndrome (see Section IV.C.1).
b. Pathology.
Hereditary isolated tumoral (or type II) hyperaldosteronism in Australia was diagnosed in 67 patients among 27 families, representing a higher prevalence than for nontumoral hyperaldosteronism (97). Because bilateral adrenal hyperplasia has been even more prevalent than unilateral tumor among those cases and because both adrenal features have been diagnosed in the same family, it is unclear whether all these families have one cause and whether the causes are truly monoclonal in the adrenal. Bilateral macronodular disease with hypercortisolism has been reported in one large family (98).
Primary (sometimes termed type II) adrenocortical tumor should be distinguished from secondary hyperfunction caused by a steroid biosynthetic defect (sometimes termed type I) with secondary adrenal stimulation by ACTH, angiotensin II, and/or hyperkalemia (36, 99).
c. Genes and loci.
Genome-wide analysis in 8 members of the largest Australian family with type II or tumoral hyperaldosteronism showed linkage to 7p22 (100).
7. Isolated carcinoid tumor
a. Expressions.
Isolated carcinoid tumor has not been reported in many members of any family. So the few familial clusters might be random. However, epidemiological analyses have suggested greater than chance association (101, 102). Most of the few reports describe families with two members expressing hindgut carcinoids. In theory, isolated foregut carcinoids (thymus, bronchus, stomach) could be an incomplete expression of MEN1 or NF1. This has not been reported. Two apparently unrelated families each had first degree relatives with bronchial carcinoid but no other features of MEN1 (103). Duodenal somatostatinoma (sometimes termed carcinoid) is rare in NF1 and, therefore, also unlikely to account for several isolated cases in a family. One family has had 3 members, each with multiple duodenal carcinoids (104).
b. Pathology.
No specific pathological features have been recognized
c. Genes and loci.
No subchromosomal locus has been mapped for familial isolated carcinoid tumors
B. Hereditary hormonal excess in one monoclonal tissue within a multiple neoplasia syndrome
1. HPT-JT
a. Expressions.
HPT-JT was not recognized until about 1990, and approximately 40 families have been reported. The features are hyperparathyroidism (90%), which includes parathyroid cancer (15%), cemento-ossifying jaw tumors (30%), renal cysts (15%), renal hamartoma [highly penetrant in one family (105)], uterine tumor (455), and Wilms tumor (rare) (70, 76). Progression to chronic renal failure is rare. Among the hereditary syndromes with hyperparathyroidism, HPT-JT stands alone in its high penetrance for parathyroid cancer. This cancer has metastasized systemically as early as age 26 (our unpublished data). Despite their high frequency of malignancy, the parathyroid glands are involved asynchronously; sometimes only a single parathyroid tumor is found at surgery (70). The parathyroid cancer may be PTH nonsecreting (106).
b. Endocrine pathology.
About 20% of parathyroid tumors in HPT-JT, whether benign or malignant, show an otherwise unusual micro- or macrocystic component. The renal lesions also are principally cystic, in the form of large cysts that are often bilateral.
c. Genes and loci.
If a family has been large enough to allow genetic linkage testing in HPT-JT, linkage has always been significant at 1q24-q32, suggesting a monogenic origin (76). The mutated gene in HPT-JT is HRPT2 (76). A small subset of families with apparently isolated hyperparathyroidism has also shown HRPT2 mutation (456). Most sporadic parathyroid cancers have HRPT2 mutation, and an HRPT2 mutation is in the germline in up to half (107).
Because the identified germline and somatic mutations all predict premature truncation or absence of the encoded protein, HRPT2 likely contributes to tumors by a loss-of-function mechanism (76). There is 1q LOH in up to half of tested parathyroid or renal hamartoma specimens (105). 1q LOH is also found in 20% of sporadic parathyroid tumors with cystic features, with somatic HRPT2 mutation in 5% of this sporadic and cystic subgroup (76, 108). Immunohistology showed loss of HRPT2 encoded parafibromin in parathyroid cancers (457, 458).
d. Molecules.
The HRPT2 gene predicts a protein (parafibromin) with unknown function. It has a weak amino-acid sequence homology to cdc73p, a yeast protein that is part of a large complex with the polymerase II transcription apparatus (109, 110). A similar complex with parafibromin has been found in human tissues (111).
2. Cowden and Proteus syndromes
a. Expressions of Cowden syndrome.
Cowden syndrome is a multiple neoplasia/hamartoma syndrome (112). The main features are fibrocystic breast disease (75% of females) with very high risk of malignancy (25–50% of females) and skin lesions (over 90%), particularly trichilemmomas (hamartomas of the hair follicle). Additional features are hamartomatous colonic polyps, early-onset uterine leiomyomata, lipomas, endometrial cancer, macrocephaly, and mental retardation.
Non-MTC occurs in up to 10%. Benign thyroid lesions including goiter are more common (50–70%). Thyrotoxicosis is not increased.
b. Expressions of Proteus syndrome.
The name Proteus is from the Greek god, who could change shape at will. Proteus syndrome includes areas of localized overgrowth, expressed as narrowly as in one digit or as extensively as hemihypertrophy, and with combinations of macrocephaly, facial asymmetry, hyperostosis, and patchy skin changes (11). The skin changes include verrucous nevi, intradermal nevi, hemangiomas, lipomas, and plantar lesions. Ovarian or testicular tumors occur in 25%. Proteus syndrome is included here because of a sometime relation to the PTEN gene (see Section IV.C.2.d).
c. Endocrine pathology of Cowden syndrome.
The thyroid cancer is usually follicular but can be papillary; this contrasts with the papillary predominance in sporadic and other hereditary non-MTC.
d. Genes and loci.
Inactivating mutation in the PTEN gene causes all or most cases of Cowden syndrome. Mutations of this gene have also been identified in Bannayan-Riley-Ruvalcaba syndrome (lipomatosis, macrocephaly, and speckled penis) and in two of nine cases of Proteus syndrome (113). The PTEN mutations in Cowden syndrome or Bannayan-Riley-Ruvalcaba are germline; Proteus syndrome is considered a mosaic disorder (see Section IV.C.2.b). Although PTEN mutations are associated with the three described phenotypes, no clear relations to genotype have emerged. Thus, all these patients can be grouped together as PTEN hamartoma-tumor syndrome.
LOH at the PTEN locus at 10q22–24 occurs in 25% of sporadic follicular thyroid tumors, but somatic mutation of PTEN there is rare, so any relation to PTEN is uncertain (114). PTEN inactivation contributes clearly to selected sporadic tumors. For example, about half of sporadic endometrial cancers and many colon cancers have somatic PTEN mutation (115, 116).
e. Molecule.
PTEN encodes pten, a dual specificity lipid phosphatase. In particular it is the 3'-phosphatase for phosphatidylinositol-3,4,5-triphosphate (PI3,4,5P) (117). PI3,4,5P binds to and regulates protein kinase B/Akt and other pleckstrin homology domain-containing regulatory enzymes (118).
3. Adenomatous polyposis of the colon (APC)
a. Expressions.
APC or familial adenomatous polyposis (FAP) is a syndrome of multiple colonic polyps with high likelihood of progression to cancer in one or more polyp (66). Other features of FAP are desmoids, congenital hypertrophy of retinal pigment epithelium, jaw cysts, sebaceous cysts, and osteomas.
Papillary thyroid cancer occurs in approximately 1–2% of cases, or about 10-fold the general population prevalence (39). There is no association with hyperthyroidism. About 90% of thyroid cancer in FAP is in females with average onset at age 25.
b. Endocrine pathology.
The thyroid tumors in FAP are usually multicentric and papillary, showing in some segments the otherwise unusual cribriform or cribriform-morular variant appearance. Cribriform histopathology is heterogeneous and includes anastomosing arches of cells and solid nests that are unencapsulated, have little stroma, and can infiltrate surrounding parenchyma (119, 120).
c. Genes and loci.
Germline intragenic mutation of APC is identifiable in up to 80% of index cases with FAP; most of the remainder have large APC deletions that are difficult to detect (66). The APC gene is on chromosome 5q and is inactivated early in colonic tumorigenesis, with biallelic inactivation being identifiable in 80% of benign or malignant tumors in FAP or similarly in sporadic colonic tumors (66). Biallelic inactivation of APC also is found in the associated thyroid tumors.
There are modest genotype/phenotype correlations. Most of the thyroid cancers arise in cases with germline mutation in the largest APC exon 15, similar to the genotype of FAP cases with congenital hypertrophy of retinal pigment epithelium (121). Other phenotypes also have distinct genotypes; cases with other extracolonic features have mutation at codons 1403–1578, and those with attenuated FAP have truncations before codon 157.
d. Molecules.
Apc is a cytoplasmic protein; it binds in vitro to at least seven categories of proteins, including ß-catenin; the latter seems central. Sporadic papillary thyroid cancer without APC mutation sometimes shows mutation in other genes of the wnt signaling pathway; in particular, it may show mutation of and/or abnormal compartmentalization of ß-catenin (122). This is particularly true also for the cribriform variant with or without APC (123). Similarly, colon cancers without APC mutation generally have ß-catenin mutation, suggesting that both genes function in the same pathway, and that disruption in either one is sufficient for an equivalent contribution to this tumorigenesis (124). ß-Catenin accumulates particularly in the nucleus with APC loss of function.
4. Werner syndrome
a. Expressions.
Werner syndrome4 is progeria or accelerated aging with impaired growth (90%), high-pitched voice (80%), early cataract (95%), skin atrophy (90%), soft tissue calcification, osteoporosis (40%), atherosclerosis, diabetes mellitus (50%), graying and loss of hair (30%), and hypogonadism (60%). Werner syndrome is a recessive disorder with greatly increased risk of selected nonepithelial neoplasias (125). They include soft tissue sarcoma, meningioma, melanoma, and osteosarcoma. These neoplasias are the leading cause of death. Thyroid cancer prevalence (10%) is increased, but thyrotoxicosis is not (126).
b. Endocrine pathology.
Among 23 cases of Werner syndrome with thyroid cancer, only 35% were papillary (127), representing a 3-fold excess of follicular histology, compared with sporadic cases. Cells from Werner syndrome have increased genome instability, seen as chromosomal deletions, reciprocal translocations, and inversions (128). No specific DNA repair system has been implicated.
c. Genes and loci.
All or most cases of Werner syndrome arise from homozygous mutation of the WRN gene at 8p21. All mutations predict loss of function, and there is no correlation of genotype with phenotype. Mutation of WRN has not been reported in somatic tissues.
d. Molecules.
WRNp sequence predicts a member of the DExH box family of DNA and RNA helicases. The protein or nucleotide interactions of WRNp are not known.
5. Li-Fraumeni (cancer family) syndrome
a. Expressions.
The Li-Fraumeni syndrome is phenotypically heterogeneous (129). Important features include bone or soft tissue sarcoma diagnosed under age 45, breast cancer, brain cancer, and leukemias.
Adrenocortical carcinoma occurs in about 1% and usually before age 14. This has been studied in a large group of children with adrenocortical tumor in Brazil. Four Brazilian families each had two to five cases with adrenocortical tumors. Average age at diagnosis of adrenal tumor was 3 yr, and most adrenal tumors were steroid hormone-secreting. About half of the children were virilized, including one fourth with hypercortisolism as well (130).
b. Endocrine pathology.
Most (72%) of the adrenocortical tumors in the Brazilian familial setting have been malignant (131).
c. Genes and loci.
The P53 gene is on chromosome 17q. Most kindreds with typical Li-Fraumeni syndrome have a recognizable inactivating germline mutation in the P53 gene. In the large Brazilian cluster of children with adrenocortical tumor, 35 of 36 had the same Arg337His P53 mutation, reflecting common ancestry (132). This mutation has not been reported in other families; thus, it seems to correlate with high penetrance for the adrenocortical tumor phenotype (133). Somatic mutation of P53 is also found in many tumors, including 70% of sporadic adrenocortical cancers (134).
A similar cancer family phenotype ("Li-Fraumeni like") with even greater excess of breast cancer occurs with heterozygous mutation of CHK2/CHEK2. The latter mutated gene can also predispose to familial isolated breast cancer or familial isolated prostate cancer (135). It has not been associated with adrenocortical tumor.
d. Molecules.
The P53 gene encodes a 393-amino acid nuclear phosphoprotein (66, 129). It belongs to a gene family with P40, P51, P63, and P73. p53 is believed to be a transcription factor that binds to a small DNA motif in the promoter of many genes.
6. Beckwith-Wiedemann syndrome
a. Expressions.
Cases have macroglossia, abdominal wall defects (including omphalocele), GH-independent macrosomia, and craniofacial dysmorphism. There is increased risk of Wilms tumor, hepatoblastoma, and rhabdomyosarcoma. Neonatal hypoglycemia occurs in 30–60% and usually resolves over 3 d (136). Adrenocortical carcinoma is occasional and can be associated with virilization. One case had bilateral pheochromocytoma (137).
b. Endocrine pathology.
There are adrenocortical cytomegaly and cysts. IGF-II is overexpressed in several related sporadic neoplasms including adrenocortical tumors and Wilms tumor (36). Neonatal hypoglycemia seems secondary to hyperinsulinism, but the cause of the hypoglycemia needs further clinical analysis (136).
c. Genes and loci.
The genetic and molecular bases have not been established. Occurrence is sporadic or autosomal dominant with variable penetrance. Mapping in families established a locus at 11p15.5 (distinct from WT1 at 11p13), which includes the IGF-II gene. Some cases show uniparental disomy across a large region of 10 Mb, which also includes the IGF-II gene. This particular background occurs in mosaics (138). This is a complex locus that also includes the H19 and p57/KIP2 genes. These three and several other genes in this locus are imprinted. Loss of maternally expressed suppressor(s) (such as p57 and/or H19), gain of paternally expressed growth promoter(s) (such as IGF-II), or even combination of the above might contribute to tumor development.
d. Molecules.
The causative genes and thus their encoded molecules remain unproved.
7. TSC
a. Expressions.
Principal features of TSC are tuber of brain cortex (93%), subependymal nodule (95%), subependymal giant cell astrocytoma (6%), cardiac rhabdomyoma (50%), renal cysts (10%), renal angiomyolipoma (50%), facial angiofibromas (85%) (formerly misnamed adenoma sebaceum), hypomelanotic macules (more than three in 80%), forehead plaques (25%), nontraumatic ungual and periungual fibroma (20%), and learning disability (50%) (11, 139, 140).
Pancreatic islet tumors or cysts are an uncommon but definite hormonal feature. They rarely oversecrete insulin. The Eker rat with a similar syndrome due to a TSC2 mutation expresses nonfunctioning pituitary tumors (141). TSC is included herein because of its occasional hormone-producing tumors and because of its overlaps (mainly nonhormonal) with other hormonal excess syndromes, particularly MEN1 (Table 2A).
|
c. Genes and loci.
TSC can arise from germline mutation in one of two genes, TSC1 on chromosome 9 or TSC2 on chromosome 16. About two thirds of cases arise as new mutations. The disease phenotype from germline mutation in either gene is indistinguishable. Most mutations of TSC1 or TSC2 predict truncation or absence of the encoded protein. Separated by only 6 bp at the 5' end of TSC2 is the otherwise unrelated PKD1 gene, cause of polycystic kidney disease. Contiguous gene syndromes can inactivate both TSC2 and PKD1, causing TSC with renal cysts and often renal compromise (144).
In general, biallelic loss of function at the locus for either TSC1 or TSC2 has been demonstrated in many of the various nonhormonal tumors of cases with TSC1 or TSC2 mutation and in some similar sporadic tumors.
d. Molecules.
TSC1 encodes hamartin, and TSC2 encodes tuberin. Hamartin and tuberin are cytoplasmic proteins that are found as a tight complex with each other. Most disease-associated missense mutations of either disrupt this complex (145). Tuberin has a region of GTPase activating protein (GAP) homology. The small GTPase Rheb (Ras homolog enriched in brain) is a direct target of tuberin GAP activity (146). Growth factor signaling through PI3 kinase and protein kinase B/Akt inhibits the GAP activity of tuberin/hamartin, resulting in Rheb activation (147).
8. Paraganglioma-gastric stromal sarcoma/Carney triad
a. Expressions.
Five families that each include two to three affected members with one or both of multifocal extraadrenal functioning paraganglioma and gastric stromal sarcoma have been proposed to have a novel autosomal dominant syndrome (148). These families were identified among a larger number of cases with the Carney triad, of which the third main feature is pulmonary chondroma. Some cases with Carney triad also have nonfunctioning adrenocortical tumor (13%), esophageal leiomyoma (9%), or other tumor less frequently (149). The full Carney triad has not otherwise occurred on a familial basis and is distinguished by a female sex preponderance. It is not clear whether the Carney triad reflects changes in gene(s) different from that/those speculated in the five familial clusters. Expressions of paraganglioma were similar to those in other paraganglioma syndromes, including occasional pheochromocytoma (see Section IV.A.3.a). There is a propensity for unusual paraganglioma locations such as the aortopulmonary body.
b. Endocrine pathology.
Pathological features of the paragangliomas are similar to those in other paraganglioma syndromes.
c. Genes and loci.
No identification or even mapping of a causative gene has been reported
C. Multiple endocrine neoplasia (MEN) syndromes
By definition, each MEN syndrome includes neoplasm in at least two different potentially hormone-secreting tissues. Each of these syndromes also includes one or more additional neoplasms. Hormonal excess is the most prominent expression for only two of these syndromes, MEN1 and MEN2 (Table 3).
|
).
b. Endocrine pathology.
An overriding feature of MEN1 at the macroscopic level is tumor multiplicity (22, 31). At the time of parathyroidectomy in MEN1, typically three or four parathyroid glands hold tumors. Similarly, gastrinomas (duodenal) in MEN1 tend to be small, submucosal, and multiple; these features, the morbidity of duodeno-pancreatic surgeries, and possibly the presence of occult metastases make gastrinomas in MEN1 usually refractory to surgical cure (151, 152).
Gastrinoma in MEN1 shows a different tissue distribution than that of normal gastrin cells (in the gastric antrum). Thus, some consider all extragastric gastrinomas to be ectopic and malignant, independent of histological grade. The less common tumors inherent in MEN1 are generally solitary, including insulinoma and pheochromocytoma. Malignancy of enteropancreatic neuroendocrine of foregut carcinoid in MEN1 is frequent. It is vascular like non-MEN1 cancers in these tissues. Malignancy in the parathyroid of MEN1 is extremely uncommon, despite multiplicity and early age of onset for the parathyroid tumors (73). There is no established precursor stage for an MEN1 tumor. However, a mouse model for MEN1 has shown striking ß-cell hyperplasia as an insulinoma precursor stage (153).
c. Genes and loci.
All typical MEN1 cases probably arise from germline mutation in the same gene. The MEN1 gene is at 11q13 and contributes to tumors mainly via a biallelic loss-of-function mechanism. Eighty percent of the germline or somatic MEN1 mutations predict truncation or absence of menin protein; the remaining 20% of germline mutations predict change of one or few codons (missense) and are also likely to cause loss of menin function. No clear genotype-phenotype correlation has been recognized; however, among 19 germline MEN1 mutations found in the minority of all families with isolated hyperparathyroidism, there has been a mild clustering of missense mutations about codons 255–413 (70, 154).
Most of the intrinsic tumors in MEN1 have biallelic MEN1 inactivation (31). This is generally suggested by 11q13 LOH. Rarely, the wild-type allele is inactivated by a small MEN1 somatic mutation that does not result in a zone of 11q13 LOH (155). 11q13 LOH has not been found in the adrenocortical tumors or thymic carcinoids that are clearly associated with MEN1 (156, 157), raising the possibility of two specific tissue types with a differing step of tumorigenesis. The adrenocortical tumors in MEN1 are generally benign, whereas the mediastinal thymic carcinoids are generally malignant.
Among related sporadic tumors, biallelic somatic mutation of the MEN1 gene is common in parathyroid adenoma (20%), insulinoma (10%), gastrinoma (20%), and bronchial carcinoid (40%); however, it is rare in anterior pituitary tumor (2%), follicular thyroid tumors (below 2%), leiomyoma (64), or small cell lung cancer (31, 158).
d. Molecules.
The MEN1 gene encodes menin, a widely expressed protein of 613 amino acids. Menin resides mainly in the nucleus and has been localized to telomeres during meiosis (159). The physiological functions of menin are not known, but menin can interact in vitro (directly or indirectly and with unknown importance) with a growing list of more than 10 proteins (160). Menin partners with the strongest suggestions for involvement in tumorigenesis are junD and a large complex, homologous to the COMPASS complex in yeast (161, 162, 163, 460).
2. MEN2A and MEN2B
a. Expressions.
MEN2A consists of MTC (95%), pheochromocytoma (50%), and hyperparathyroidism (30%). MEN2B has the features of MEN2A, plus intestinal ganglioneuromas and the mucosal neuroma phenotype (90%), but minus hyperparathyroidism (50). Most aspects of FMTC are considered separately (see Section IV.A.4). Each of the three is a variant of MEN2. The relative prevalences of index cases are MEN2A>>MEN2B>FMTC (50, 93, 164). Late dissemination of MTC can be to lungs, liver, and/or brain. The relative ordering of MTC for lower age of onset and greater aggressiveness is highly syndrome-dependent: MEN2B>MEN2A>FMTC. Earliest age of known dissemination of MTC in a variant of MEN2 has been central in determining the earliest age for suggesting prophylactic total thyroidectomy for carriers. Metastases of MTC have been recognized as early as the age of 1 yr in MEN2B, as early as age 5 yr in MEN2A, and rarely before age 15 yr in FMTC. Pheochromocytoma was formerly a major source of morbidity, similarly in MEN2A and MEN2B. This is no longer true because of prospective monitoring and improved pharmacological managements, particularly around surgery. Pheochromocytoma causes more symptoms in MEN2 than in VHL; this correlates with higher tumor expression of tyrosine hydroxylase and more tissue stores of catecholamines in MEN2 (165).
Among the three main hormonal excess states of MEN2A, hyperparathyroidism causes the least morbidity. Parathyroid involvement in MEN2A can reach 50% or more by age 70 yr and can be in multiple glands (166, 167). Parathyroid tumors are more asynchronous in MEN2A than in MEN1.
b. Endocrine pathology.
C cell neoplasia begins as multifocal hyperplasia and then becomes micronodular. C cell cancer can also be multifocal in MEN2 (168). In fact, it has been suggested that monoclonality within C cell hyperfunction is present at the stage of hyperplasia and might even begin at embryonic stages before lobation of the thyroid (169). Eventual spread is to regional lymph nodes.
Pheochromocytoma in MEN2 also begins as a hyperplastic stage before progression to mono- or oligoclonality as evidenced by biallelic inactivation at the VHL locus and by LOH at 1p (170); a similar implication of additional loci has not been seen with pheochromocytoma in VHL (171). This tumor often remains histologically benign but sometimes aggressive metabolically. Monoclonality of the parathyroid tumor in MEN2A has not been documented.
c. Genes and loci.
In the recent past, calcitonin assay after challenge with pentagastrin was used as a sensitive and semiquantitative index for MTC, C cell hyperplasia, and thus for ascertainment of the MEN2 carrier state. MEN2 carrier ascertainment by serum calcitonin protocols has been largely replaced by the more effective sequencing for germline RET mutation (50, 93).
The RET gene is at chromosome 10q11.2, and one of its germline mutations is detectable in 98% of index cases with an MEN2 variant (50, 93). In fact, the RET locus has never been excluded by chromosome 10 haplotyping in any MEN2 family. The Cys634 codon accounts for 80% of germline mutations in MEN2A, whereas the Met918 codon accounts for 95% of germline mutations in MEN2B (Fig. 1). RET mutations are found in somatic tissues of sporadic MTC or sporadic pheochromocytoma, but with a very different frequency distribution of mutated RET codons. Met918 is the most frequent somatic RET mutation in sporadic tumors and accounts for 25% of the RET codon mutations in these (50). Somatic gain-of-function RET missense mutation is found in about one fourth of sporadic MTC and one tenth of sporadic pheochromocytomas. It has not been found in sporadic parathyroid adenoma. RET-PTC, a different mechanism of somatic RET gain-of-function mutation, leading to a ret fusion protein, is common in sporadic nonmedullary papillary thyroid cancer (see Section IV.C.2.d).
d. Molecules.
Ret protein with four tandemly repeated extracellular cadhedrin-like domains is a transmembrane tyrosine kinase, part of a large family that includes the C-MET oncogene. Upstream and downstream components of the normal ret pathway have begun to be clarified (see Section VII.G.2) (18). The ret system is functional during organogenesis in intestinal ganglia and the kidney. This correlates well with loss-of-function mutations in RET that are a major cause for familial Hirschsprung disease (intestinal aganglionosis); it is thus an interesting paradox that occasional cases of MEN2A (with presumed RET gain-of-function) also express Hirschsprung disease (18).
There is a striking correlation of RET genotype and MEN2 phenotype (Fig. 1). Most germline RET mutations in MEN2A are at cysteine residues in the cysteine-rich extracellular domain about cysteine-634. Most MEN2B RET germline mutations are at methionine-918 at its cytoplasmic tyrosine kinase catalytic site. RET germline mutation in FMTC occurs in a zone that spans both of those domains but shows an extra location, almost specific to FMTC at codons 768–844. All MEN2-associated RET mutations predict missense, and all are believed to cause ret gain of function (18). The extracellular mutations probably cause gain of ret function by promoting ret dimerization; the MEN2B-associated mutations in the cytoplasmic tyrosine kinase domain activate the tyrosine kinase directly. Tyrosine kinase inhibitors have already been targeted to certain cancer genes, and the RET gene is under exploration for this (172).
Somatic but not germline RET mutation is common in papillary non-MTC. Most sporadic papillary (and some Hürthle or follicular) thyrocyte tumors have a 5' (i.e. amino-terminal) fusion of RET with one of eight other genes. The fusion genes are termed RET-PTC1 through RET-PTC8 (173, 174). Progression to poorly differentiated thyroid cancer is rare with RET-PTC (175). Each of these eight gene fusions can have any among three consequences that favor RET gain of function: 1) a new promoter drives ectopic expression in thyrocytes; 2) abnormal compartmentalization in cytoplasm is favored by loss of the transmembrane domain; and 3) the fusion protein has a new N terminus, favoring homodimerization.
3. VHL
a. Expressions.
VHL is a highly penetrant disorder with apparent age of onset that varies with the family and the thoroughness of evaluations (11, 176). The main features of VHL are nonhormonal (Table 3): clear cell (sometimes with occasional granular cell) renal cancer (40%), renal cysts (35%), cerebellar and spinal hemangioblastoma (60%), papillary cystadenoma of the epididymus or broad ligament (15%), and retinal angioma (60%). Retinal angioma is typically the feature with earliest age of onset during thorough prospective screening. The renal carcinomas are multifocal and often will recur with current management by conservative or kidney-sparing resection.
The hormonal features of VHL are pheochromocytoma (20%) and pancreatic islet tumor (10%). Pheochromocytoma clusters only in selected VHL families, leading to their classification as VHL type II; furthermore, those rare families with VHL mutation and with only pheochromocytoma are classified as VHL type IIC (176). Pheochromocytoma in VHL may become symptomatic during childhood. The catecholamine-related biochemical profile differs between pheochromocytomas of VHL and MEN2A (see Section IV.C.2.a). Pancreatic islet tumor in VHL as in TSC does not usually oversecrete hormones but presents as a mass.
b. Endocrine pathology.
The nontumorous adrenal medulla in VHL, unlike in MEN2, does not show chromaffin hyperplasia, and it shows a characteristic amphophilic or clear cytoplasmic morphology (177). The pancreatic islet tumors are generally benign and multiple. Islet histology is micro- or macrocystic, serous, or adenomatous. Many islet tumors or cysts in VHL immunostain for insulin but rarely, if ever, cause hypoglycemia (178). The hormonal and nonhormonal tumors of VHL have either large vessels or a prominent vascular component; this seems related to their overexpression of VEGF (179).
c. Genes and loci.
A VHL germline mutation is identifiable in virtually all VHL
