Clonality Studies in the Analysis of Adrenal Medullary Proliferations: Application Principles and Limitations.

Clonality remains as the hallmark of neoplasms. A dual genetic approach using markers nonrelated (e.g., X-chromosome inactivation assays) and related to the malignant transformation (such as loss of heterozygosity analyses of tumor-suppressor genes) would provide useful clonality information from early and advanced tumor stages, respectively. Tumor progression and clonal selection would result in genetic instability and heterogeneous expression of those molecular markers related to the malignant pathway. Therefore, only the coexistence of multiple genetic abnormalities would support the clonal nature as an expression of convergent cell selection. Considering those facts, the currently available evidence on tumorigenesis and clonality in the adrenal medulla can be summarized as follows: 1. Multistep tumorigenesis defines the evolution of pheochromocytomas, as evidenced by the presence of several genetic alterations. 2. Both the significant association of nonrandom genetic alterations (specially 1p and 22q interstitial deletions) and the topographic accumulation of genetic deletions at the peripheral tumor compartment support a convergent clone selection for these neoplasms. 3. Although many genetic loci show nonrandom abnormalities, the most frequently involved locates on chromosome 1p regardless of genetic tumor background (sporadic or inherited predisposition). 4. Most pheochromocytomas should begin as monoclonal proliferations that do not always correlate with histopathologic features, particularly in inherited tumor syndromes. 5. Early histopathologic stages, described as adrenal medullary hyperplasias, are defined by hyperproliferative features in animal models and monoclonal patterns in the adrenal nodules from patients with MEN-2a.


Introduction
Although many molecular features have been described in order to characterize neoplasms better, its definition is still descriptive. All these f~atures try to distinguish neoplasms from nontumoral conditions and malignant from benign neoplasms as well. Malignant tumors (specially carcino-mas) evolve through continuous and additive changes, both at genetic and phenotypic levels, ending in the complete malignant phenotype. The most widely accepted theory proposes a clonal origin for them, in contrast to the heterogeneous composition in most normal tissues [1,2]. Therefore, monoclonal cell populations 301 arisen from heterogeneous normal tissues strongly suggest a ~eoplastic nature. Different and variable success has been reported for several markers, but clonality still remains as the halhnark of neoplasms. Befbre any specific reference to clonality in pheochromocytomas and related lesions of the adrenal medulla, some general biologic considerations would help us to understand the strength and weakness of molecular techniques in the analysis of clonality. Those considerations will focus on the relationship among cellular kinetics, neoplastic transfbrmation, and clonality. A brief review of the morphologic criteria of pheochromocytomas (PCC) and adrenal medullary hyperplasia will follow, and finally, the current information on clonality in the analysis of adrenal medullary proliferations will be covered.

General Considerations on Malignant Transformation and Clonality
The term clonality has been broadly used, and it is related to the presence of cellular clones in tissues, although that terminology can be applied at different organizational levels. Those clones represent the groups of genetically identical copies of a single progenitor. The strict application of that definition would require the sequence of all DNA in order to exclude any polymorphism, but that approach is not viable. Therefore, different markers have been proposed to obtain such information using more or less complicated and simple techniques. In any case, the presence of groups of genetically identical cells must be related to proliferative advantages of those cells over the remaining. For that reason, a close relationship can be drawn between clona.lib ' and kinetics features. The expression of those features is variable fi:om the conventional histologic level to the molecular one.
The current definition of neoplasm is descriptive and closely related to clonalit3: Neoplasms are defined as abnormal and self-maintained proliferations, resulting in the presence ofnew tissues. Those new tissues should come from a group of cells sharing a particular and common advantage. Therefore, a clear relationship between both concepts (neoplasm and clone) is present. Its application to everyday practice ofpatholog7 would also contribute to explaining the development of diagnostic criteria of neoplasm in many areas of surgical pathology. Just as an example, the diagnosis of malignant lymphomas, especially at the extranodal level, requires the presence of a monotonous proliferation of lymphoid cells that substitutes the normal tissue architecture and results in a mass. Those criteria represent, at the histologic level, the morphologic evidence of an expansive growth of neoplastic cells. Sometimes, those cellular expansions share specific phenowpes (cTtoplasmic or surface markers, and so forth), which allow' their identification. Such an approach would require quantification of its expression in order to distinguish the abnormal proliferation from its normal counterpart, e.g., light-chain restriction of immunoglobulin in lymphoid neoplasms or the flow cytometric analysis ofsurt~ce lymphoid. markers in hematologic malignancies.
On the other hand, the neoplastic transformation is a multistep process, resulting in a complex mixture of tissues with heterogeneous genetic background [1.]. The results of every molecular test will depend on the timing between the evaluated biologic fbature and the specific marker (Fig. 1). Simultaneously, the sensitivity and specificity of each molecular marker used to test any biologic feature, such as clonality, A @ t t

X-Chromosome Inactivation
Xp ~ ~Xm  1. Timing between the clonal signal stimulus and the X-chromosome inactivation (XCl). The interpretation of any test result (in this case XCI) will depend on the temporal relationship between the marker used (methylation of inactive alleles on X chromosome) and the biologic feature evaluated (in this case clonatity). Normally XCI takes place early before any clonal signal (labelled + in the figure), ending in monoclonal patterns (A). Those situations where the clonal signal precedes the XCI would eventually result in polyclonal patterns, even for monoclonal proliferations ("pseudopolyclonal", B).
should be checked with samples of different and known genetic constitution, In addition, those molecular markers of clonality should be evaluated in relation to two main ~atures, cellular kinetics and the neoplastic pathway. Any marker related to tile malignant pathway will provide positive information only if that target participates in the molecular pathway of a given tumor. The current evidence suggests that malignant tumors result from the accumulation of multiple genetic events rather than from any specific sequence of genetic changes [3][4][5]. Some mutations are more important than others, because they facilitate additional phenotypic or genotypic changes. This reason explains certain trends in the order of mutation production, although significant variation i~br them is the role [6]. The lack of such a sequence and tile unpredictability of genetic changes in malignancies preclude the extensive clinical use of those genetic markers with diagnostic and prognostic purposes, Finally, the timing between the clonal expansion and the presence and expression of malignant features is unknown. All these limitations should be considered to achieve reliable and useful information from any marker. The final expression of any neoplasm depends on the progressive and convergent selection of cell populations, but several factors should be taken into consideration. On one hand, that selection will determine tumor progression and cellular heterogeneity. On the other hand, cellular selection is the expression of cell kinetics [17]. Considered from the genetic point of view, neoplasms can be characterized by multiple genetic aberrations, from gross to point mutations. However, those genetic aberrations should be "fixed" on the ttansformed cell before ending in a fully established malignancy. Any nonlethal genetic aberration will be "cell-fixed" if both its transmission to descendant cells is warranted and it is able to bypass tile cell-repairing systems. Otherwise, any genetic alteration leading to cellular aging, differentiation, or activation of the apoptotic pathway will be nontumor-productive (Fig. 2). This process determines a complex network where a potential genetic marker will be usefhl when associated with kinetic advantages responsible f?)r cellular outgrowths.
It is out of the scope ot: this article to provide detailed descriptions of those markers used for testing both clonality and cell kinetic I17-9]. Essentiall> two types of molecular markers provide reliable results on clonality, classified according to its chromosomal location in X-linked and non-X-linked assays. Those markers independent of the malignant transformation pathway provide more general information, especially fiom the early neoplastic stages. Among them, the assays based on the ralldom inactivation of one X-chromosome in females are the most extensively used to test clonality [9]. These tests target the differential expression of alleles of X-chromosome-linked genes (mt~XlA or protein analyses) or the differential DNA methylation on active (nonmethylated) and inactive (methylated) alleles, Currently, tile most productive marker of this sort is the human androgen receptor gene (HUMARA) owing to its high percentage ofheterozygosity in the general population (around 90%) related to the presence of a highly polymorphic trinucleotide repeat region. That polymorphic region also shows a predictable methylation pattern on sequences recognized by methylationsensitive restriction endonucleases [10], allowing allele differentiation. In i~ict, tile information on clonality provided by both X-linked and non-X-linked assays is complementary and nonmutually exclusive, because tile results from X-linked Currently, neoplasms can be considered as genetic diseases (resulted from genetic damages), which become clinically detectable after clone selection and cellular proliferation, The induction of any neoplasm requires special features for the genetic events involved in the malignant transformation. Those events (black dots) have to be resistant to the cell-repairing systems (dotted arrows), accumulated cooperatively in the cells, efficiently transmitted to descendant cells, and nonlethal (that is able to induce apoptosis). The clonal expansion of genetically damaged cells under those conditions will result in a neoplastic growth, assays are based on genetic features usually nonrelated m the malignant transi~Jrmatioil, The inactivation of one X-ch mmosome in females is an early event during embryogenesis, described in embryos of 16-30 cells [11]. ']M'o main consequences can be drawn, with boti~ biologic and technical implications. First, some tissues would show the same allele inactivated and a monoclonaJ pattern, avoiding any positive conclusion from a tumor arisen on it (nonins case) [2,7}. This pattern is variable from patient to patient and even from tissue to tissue in the same patient. For that reason, the so-called Lyonization ratio, reliable and appropriate results s any tumor or precancerous tissue samples must require embryologically-related controls [12,1311. Second, most of somatic genetic alterations described in neoplasms will take place on cells showing one X~chmmosome inactivated. Those genetic events associated with ld netic advantages and cellular overgrowths will probably determine a monoclonal methylation pattern of the X-chromosome if they trigger one cell or a small group of neighboring cells (Fig. 1).
Those tests based on other somatic markers potentially involved in the malignant pathway give information on the transformation mechanism. Any nonrandora genetic alteration (X-linked or not) present in most of the tumor cells will provide useful information on clonal expan-

Volume 9, Number 4
Winter 1998 sion, even those as small as point mutations. Those point mutations, regardless of their location on exons or introns (the latter, also referred to as single nucleotide polymorphism, has no functional expression at in RNA and protein levels), would be related to kinetic advantages and, therefore, have been proven useful to test clonal expansions in tumors. However, only those tumors expressing that particular marker will be suitable for l hat analysis. Addi donall),, the ratio ot:positive results will be necessarily related with both time-independent and time-dependent features ofeach tumor, e.g., histologic type, location, grade, or stage. ']'hose variables can be "confusi~g factors" if they are not adequately evaluated. Owing to the absence of an established timing for a particular genetic event in the transformation pathway, variably long gaps of test negativity would be demonstrated tot a particular molecular marker in a given tumor (Fig. 3). That would cover the time between tumor initiation and assay conversion, giving inconclusive results. There will not be any gap time f'or such conversion if the tested molecular marker represents the first mutation event in the natural history of a given neoplasm.
In that case, the test results would be identical to those provided by markers unrelated to the malignant transformation pathway, e.g., X-chromosome-inactivation assays. In conclusion, only:' multifi~ctorial approaches considering different markers simultaneously would appropriatdy provide clonality information in tumor cmyditions.

Pheochromocytomas and Adrenal Medullary Hyperplasias: Basic Pathology
In general, the mort~hologic features of. endocrine tumors barely correlate with biologic behavior. Many clinicopathologic studies make the point that cytologic features, such as nuclear pleomorphism, hyperc/lromatism, or increased nuclear/ cytoplasmic ratios, are not necessarily related with aggressive outcome. Sometime, only the presence of metastatic tumor growths enable us to establish a malignan t diagnosis [14]. Therefore, in addition, histopathological studies are requhed to improve both diagnosis and prognosis of these neoplasms. Other approaches trying to detect the minor at early steps (precancerous conditions or incipient neoplasms) go in a parallel direction to improve patient prognosis. Pheochromocytomas (PC(~) fulfill all these general rules.
The histopathologic differentiation of hyperplastic and neoplastic conditions in the adrenal medulla is mainly arbitrary [ 1 ~, 15]. The first d iffcrentiating criterion is the increase in the adrenal weight, as an expression of the increased number of chromaffin cells, with expansion of the medullary compartment into areas of the gland where it is not normally present. Mor phologicall> hyperplastic conditions have been classified according to their growth pattern as cliff:use or nodular [16]. Additionally, no specific and consistent criteria will distinguish big nodules with in a muhinodular background (hyperplastic condition) from small PCC (neoplastic conditiot~). Some reports propose the nodule size as the main differentiating criterion and 1 cm as the threshold [17]. That size has been suggested as the criterion becau.se it is the detection limit of most image diagnostic systems and represents the smallest size clinically detectable. That tumor size was also selected because the smallest PCC in the first series from the Armed Forces Institute of Pathology had l--cm diameter [15]. No biologic reasons are provided t~)r that distinction. On the other hand, the krmwledge available for ad renal medullary hyperplasia comes from inherited condi- The natural history of neoplasms is a linear temporal expression of the convergent and endless selection process of cell clones in tumors, The clone selection results from the interaction between genetic changes (mutational events) and individual susceptibility. Progressive and cooperative changes should accumulate in cells before they are expressing a fully neoplastic phenotype, The point of irreversibility in that sequence also separates the chronologic evolution from preneoplastic into neoplastic conditions. On that background, the clonality markers will give useful information to depict the neoplastic natural history. Any particular marker will give information concerning clonality if it is involved in the malignant transformation process. Likewise, it will be concordant with XCI assays only if it represents the first mutational genetic event. All other situations will determine negative results for the marker even for already clonal proliferations (negative test gap).
tions, especially those related to MEN-2, where quantitative analyses demonstrated utility to distinguish medullary hyperplasias from normal adrenal medulla [16]. Cytologic and architectural features have showed no significant and reliable differences useful for that task [t5].
PCCs have been described as sporadic neoplasms or associated to other tumors as part of inherited cancer syndromes, mainly multiple endocrine neoplasia type 2 (MEN-2), neurofibromatosis, or yon Hippel-Lindau disease [18]. Clinically and histopathologically, the situation is di(:f~er-ent for both conditions [15]. The absence of either a known genetic background or a reliable screening method precludes the systematic detection of early sporadic tumors. Therd~)re, most of those sporadic tumors are detected in more advanced stages, although only about 10% shows malignant behavior. Additionally, no preneoplastic changes are usually associated with them. The opposite situation characterizes the inherited PCC: they are usually detected earlier owing to genetic and clinical screening, and medullary hyperplasia, either diffilse or nodular, is also present.
Clinicopathologic studies mainly define benign and malignant PCCs. Currently, a bona fide diagnosis of malignant PCC requires the demonstration of" metastases, defined by the presence of tumor overgrowths in sites where chmmaffin tissue is not normally present [I5]. Those organs include lymph nodes, liver, and bone [14]. The presence of nonneoplastic chromaf'fin tissue in other organs (as a physiologic finding or as a part of hyperplastic conditions) avoids the establishment of malig nant diagnoses based merely on the histopathologic evidence of such tissue. The demonstration of truly metastatic outgrowths would require additional studies in order to prove clonaI identity in both primary a~d metastatic tumor [19]. Relatively clear patterns arc usually shown at both ends of the pathologic spectrum (benign and malignant tumors) [20]. A slightly different situation would be expected in borderline conditions, such as locally invasive tumors [21,22]. Those tumors are characterized by variable invasion ofperiadrenal sof't tissue, with no evidence of distant metastatic growths. Its association with lymph node metastases needs to be defined and the long-term patient outcome is unknown [1511. Other parameters are required to obtain a full understanding on the biologic behavior of these particular tumors.

Evidences of Clonality Based on X-Chromosome Inactivation Analysis
lh'eliminary results revealed monoclonal methylation pattern in the alleles of the androgene receptor gene (HUMARA assay of clonality) in 87% of informative cases fiom a series of sporadic PCCs [22]..After microdissection, at least two different samples were analFzed in each neoplasm, from the peripheral and the internal tumor areas, respectively. Both tumor areas provided concordant methylation patterns, consistent with a monoclonal origin for this series of cases. The remaining 13% of informative PCCs showed balanced methylation of both t-{UMA1L~ alleles, pointing to a polyclonal tumor growth. The latter group of cases corresponded to locally invasive PCCs (as defined by periadrenal sot} tissue invasion) with high cell turnover rate (both MIB-1 labeling index and in situ end-labeling index received the highest scores for this series) [21.]. Histologically, stromal overgrowth with a prominent smooth muscle differentiation (positive immunoexpression of smooth muscle type actin and desmin) was revealed by dissecting the neoplastic celt nests. DiftCerential microdissection from several rumor locations always confirmed the initial results. Excluding the possibili F of significant contarnination with host normal cells by careful and repeated sampiing with microdissection, those findings could be the result of (1) true polyclonal tumors, which, because of an unfinished process of neoplastic cell selection, still show coexistent cell clones descended t~rom different progenitor, or (2) an abnormal methylation of the H UMAtL,~t locus during tumor progression (hypermethylation of the active allele) expressing pseudopolyclonal patterns in true monoclonal neoplasms.
Another method used to investigate clonality, also based on the X-chromosome inactivation, depends on the mosaicism of protein expression in the normal tissues from heterozygous women showing two isoforms of the enzyme glucose-6-phosphate dehydrogenase (G6PD) [33][34][35]. The original report ofclonatity in medullary thyroid carcinomas and PCCs associated with multiple endocrine neoplasia relies on this technique [33,34]. The authors reported the presence of only one isoenzyme in tumor tissue from patients proven to be heterozygous for that marker and concluded that the initial mutation produces multiple clones of defbctive cells. Thereafter, each tumor arises as a final mutation in one clone of these ceils. Essentially, the proposed hypothesis claims for an initial polyclonal and nonselected prolit!?ration ofdelective cells, which then turns monoclonal through a process of cellular selection [36]. The histopathologic features in each stage have not been established, although theoretically adrenal medullary hyperplasia and PCC would represent the preneoplastic and neoplastic conditions, respectively. Following the same methodology described for sporadic PCCs, HUMARA clonality assay was also performed with PCCs from members of an MEN-2a family [22]. The patients developed PCCs with no special histologic features, arisen on multinodular medullary hyperplasias. All except one tumor nodule revealed monoclonal methCation patterns, regardless of their sizes. The internodular adrenal medulla was required to show polyclonal methylation pattern in order to consider the case informative (criteria ofc~e inclusion). Therefore, no conclusions could be obtained from the adrenal medulla with diffuse growth pattern. These results support the multistep tumorigenesis in the adrenal medulla of patients with MEN-2a [16,17,26]. An initial polyclonal stage would progressively evolve into monoclonal cell growths as the result of clone selection [6,36]. On the other hand, they also question the validity of nodule size as diagnostic criteria to distinguish hyperplasias from neoplasms. "Ihmor (or nodule) size is a time-dependent parameter and revealed a low specificity for case stratification by clonal pattern. That finding can-not be considered surprising for neoplastic processes with early transformation events, as those reported in most inherited tumor syndromes [23,24].

Evidence of Clonality Based on Tumor Genetic Alterations and Multistep Tumorigenesis
Although it is generally accepted the clonal evolution takes place in neoplastic transformation [1,6], some clonality studies have reported controversial resuhs in early neoplasms and preneoplastic conditions of endocrine organs [37]. Several studies essentially confirm the multistep hypothesis for PCC tumorigenesis. Although the genetic mechanism of tumorigenesis has been refbrred to as dif}brent in pheochromocwtomas and extra-adrenal pamgangliomas [30], the same genetic targets seem to be involved in both sporadic and familial PCCs [26,29]. The general considerations mentioned above to analyze the malignant transfbrmation and clonality [7] lead us to conclude that all these reports only provide information on clonal expansions in pheochromocytomas. The genetic homogeneity for a given marker would point to a kinetic advantage, provided by the marker itself or linked to it, that represents the basic mechanism of cell selection and tumor progression [1,6]. However, this isolated finding does not prove that all turnor cells come from the same progenitor and they are truly monoclonal. The same mutagenic event can affect cells from different progenitors (Fig. 4). Therefore, the descendant cells would share the same genetic abnor-maliF (homogeneous pattern for a particular molecular marker, as expression of its clonal advan rage and expansion), although they may have distinctive genetic background (different progenitors and true polyclonal origin).
Several tumor-suppressor genes have been demonstrated to play an important role in the above-mentioned clonal expansions of adrenal medullary neoplasms. As previously shown for other neoplasms, the tumor initiation and/or progression in PCCs involves multiple genes, mainly located by LOH analyses on chromosomes lp, 3p, 17p, and 22q [26,[29][30][31]. Some oi:these markers have also revealed significant association with clinicopathologic parameters, such as tumor volume (in the case of DNA ddetkms located on 1 p, 3p, and 17p) [26] or with a distinctive transformation pathway (i.e.. lp34-36 and 3p25 deletions have been found in 45 and 56%, respectively, of PCCs, but not in extra-adrenal paragangliomas, whereas 3p21 deletions have been described in 50% of extra-adrenal paragangliomas, but not in PCCs) [30]. The relative incidence of each genetic alteration is quite variable from series to series, probably in relation to the limited number of cases analyzed. However, all series agree to show l p deletions as the most frequent genetic f'mding, although no reference is available about its timing compared with other genetic abnormalities. That information would be especially valuable for the comparison between PCCs and adrenal medullary hyperplasias in order to determine its real nature either as an initiation-related genetic event or as a consequence of tumor progression.
On the other hand, the inherent genetic instability associated with neoplasms would explain the coexistence of several genetic abnormalities [1,6]. In that way, a significant association between interstitial deletions on 22q and 1 p has been reported [29]. That finding suggests that the inactivation of multiple tumor-suppressor genes is required for PCCs development and progresskm (m ultistep tumorigenesis). Considering tumor heterogeneity, an additional conclusion arises from that association. From the statistical point of view, homogeneous genetic aherations would show a low probability for the random association of two or more molecular markers in tumor cell populations from difi ferent origins (polyclonal tissues). Previous reports on LOH analyses, using Southern blot hybridization [38] and PCR-based techniques [39], have shown random and nonmmor-related DNA deletions in 4-20% of normal tissues [40][41][42]. These percentages should be considered {tom two perspectives: (1) as a potential "confusing factor" in the analysis of case series, and (2) as the limiting threshold in order to interpret appropriately the biologic significance of any association between DNA deletions and minor conditions. Heterogeneity is evident among histologically similar neoplasms from different patients (intertumor heterogeneity) and among different cells of the same neoplasm at a single time (intratumor heterogeneity), as well as at different points in time (tumor progression). Considering that variabili9~, intratumor heterogeneity for I.OH of multiple gene loci can be exploited as biparametric markers for the analysis of clonal selection in tumor progression that LOH geneity can be expression of either selective tumor evolution or simple passive byproduct of other mechanisms, such as genetic instability. The association of multiple genetic alterations would become statistically less probable as the number of molecular markers increases [43]. Then the random association of those markers would be better explained by a convergent selective process ending in the presence of a dominant clone [1,6]. ']'hat selection mechanism of cell clones with relatively homogeneous genetic constitution would represent the best approach currently available to define clonality based on tumor genetic markers. From this point of view, clonality could be considered the key element to understanding the endless biologic process of initiation and progression of tumors (cause and consequence at the same time) that also determines the otherwise linear natural history of neoplasms (Fig. 3).
Some experimental results (personal observations) support this point of view, based on the analysis of five polymorphic DNA regions (microsatellites) located on introns of four tumor-suppressor genes (p53, Rb, WT1, and NF1). Heterogeneous DNA loss of wild-type alleles (LOH analyses) was revealed in informative sporadic I?CCs, involving p53 in 45%, Rb in 25%, WT1 in 44%, and NF1 in 50%. The comparative study of peripheral and internal tumor areas confirmed an increased accumulation of genetic deletions at the peripheral tumor compartment, expression of both tumor progression, and multistep tumorigenesis. At that peripheral level, two or more genes showed LOH in agreement with the genetic instability of neoplasms. The low probability for random and simultaneous DNA deletion in nontransformed cell populations point to clone selection, ending in outgrowths of the "selected" clone, which becomes dominant and so-called monoclonal. The final tumor picture as monoclonal proliferation thus results fiom a selective process on a genetically heterogeneous cell population (Fig. 4), Other factors also contribute to neoplastic development in endocrine organs. Somatic mutations of G-protein genes result in the constitutive activation of G-proteins and in an overall increase of the endocrine function. Along with the fimctional enhancement, proliferative advantages related to the trophic eft}ect associated with the hormonal action have also been reported in these conditions [19]. That reason would contribute to explain the fiequent association be~,een overgrowth and hyperiimction of endocrine organs. G-proteins represent a key element of the intracellular signal transduction linking the exrracellular ligands and the final cellular response. The active signal transduction. normally present in most functional endocrine systems would explain the high sensitivity-of those organs to abnormalities in that central pathway of signal transduction. That proliferative advantage would play an important role in the kinetic evolution and progressive selection resulting in cellular transformation in those disorders. Activating mutations at codon 201 of gsp have been fbund in different endocrine disorders, both neoplastic (pituitary adenomas, follicular thyroid adenomas, parathyroid adenomas, chemodectomas) and nonneoplastic (parathyroid or adrenal cortical hyperplasias) [I9]. The same activating mutation has also been reported in primary and metastatic PCCs as well as in extraadrenal paraganglioma from patients found to be wild-type at the germline DNA (extracted from leukocytes). That presence of concordant G-protein gene mutations in such a wide range of endocrine conditions and in difi~rent endocrine disorders in the same patient is consistent with a common underlying etiology.

Tumor Susceptibility and Clonality in PCC of Inherited Cancer Syndromes
The basic molecular mechanism is the same for both sporadic and inherited PCCs. However, the genetic background provides some valuable insights to understanding the general transformation process and the clonal evolution of tumors. The inherited cancer syndromes are essentially characterized by germline genetic abnormalities in certain DNA targets, frequently point mutations. Those abnormalities used to be activating point mutations targeting proto-oncogenes, such as vet in MEN-2 syndromes [32], or inactivating point mutations on tumor-suppressor genes, e.g. yon Hippel-Lindau disease [18,28] or neurofibromatosis 1 [251]. In both cases (activating or inactivating mutations), the basic mechanism for the clonal expansion is the proliferative advantage provided by an appropriate genetic network (usually as additional somatic alterations on different genes) along with the increased susceptibility" to neoplastic transformation (related with the germline mutation). Although a similar genetic background defines a given syndrome, differen t transfbrmation pathways have been described in association with each tumor t>~e and location. Therefore, variable relative frequencies of each genetic marker have been reported, i.e., using restriction fi:agment-length polymorphism analysis, lp LOH has been found in all PCCs associated with MEN-2 syndromes (67% including sporadic and Von Hippd-Lindau PCCs), whereas it was detected in only 13% of MEN-2-related medullary thyroid carcinomas [26]. The germline abnormality provides an increased susceptibility trbr tumor development that modulated by the cellular environment, will only end in neoplasms if multiple genetic loci are also involved (multistep tumorigenesis). Regardless of the predisposing genetic factor, chromosome lp deletions are again the most frequently found molecular markers, as mentioned for sporadic I?CCs (see' Evidence of CIonality Based on "Ihmor Genetic Alterations and Multistep "Ihmorigenesis) [26,29,31 ].
The genetic background also modulates and determines the molecular mechanism involved in the transfbrmation process. The most fi:equent inactivation mechanism of tumor-suppressor genes in PCCs involves both an inactivating point mutation in one allele (missense or nonsense mutations usually present as a germline defect) and the loss of the wild-type allele [26,2711, analogous with the model proposed ~br different inherited malignant tumors 1123, 24,41,42,44,45]. However, additional inactivating mechanisms should be considered when no DNA loss of the wild-type allele is revealed. Such mechanisnas include intragenic somatic mutations in the wildtype allele (also called homozygous inactivation) and hypermethylation of the tumor-suppressor gene as described for yon I-lippel-I_.indau gene [128,46,4711. On the other edge of the spectrum, several activating mutations in the ret protooncogene characterize different types of MEN-2 syndromes and familial medulla~ thyroid carcinomas (FMTC) [48]. Recentl> a mutation of ret codon 768 in exon 13 was found to segregate with the FMTC phenotype (multiple cases of medullary thyroid carcinomas or C-cell hyperplasia), but not with the adrenal medullary hyperplasia in a large multigenerational ~hmily [32]. Those preliminary data should provide information on additional factors responsible for the development of specific tumor types. However, these results should be taken with caution because of the short number of cases analyzed; in that particular series, only two individuals revealed isolated adrenal medullary hyperplasia.

Clonality and Proliferation Features in Adrenal Medullary Hyperplasias
As mentioned above, there is no general agreement on the morphologic criteria to distinguish neoplastic (incipient pheochromocytomas) from nonneoplastic proliferations (hyperplasias) in the adrenal medulla 111511. On the other hand, most of our knowledge on adrenal medullary hyperplasias comes from a restrictive group of pathologic conditions, such as experi mental pathology (in vitro and in vivo studies) and, to a lesser degree, from human pathology (in the case of inherited cancer syndromes). Those factors make it more difficult to obtain reliable and clinically relevant conclusions from the molecular analyses, useful for general application. Although no established conclusions are available at the molecular level, the proliferative response of the adrenal medulla is better known, at least under specific and controlled conditions 114:9-5211.
Current evidences suggests that chromaffin cell proliferation in adult rats is regulated by a combination of hormonal and neurogenic signals [49,5(/i]. The adrenal medulla is innervated by several nerve fibers, which stimulate the secretion of catecholamines. Therefore, after chronic adrninistration of a wide variety of catecholamine-depleting pharmacologic agents, such as the antihypertensive agent reserpine, the releasing of negative feedback Winter 1998 controls increases neurogenic stimulation on chromaffin cells [50]. The reflexively increased neurogenic stimulation of chromaffin cells is intended to meet the physiological needs of catecholamine synthesis, and would result in adrenal medullary hyperplasia and neoplasia. Catecholamine depletion would be compensated through mechanisms that normally adjust cell number, increasing the cell turnover rate and ending in chromafl]n cell outgrowths [50,52]. Many strains of rats develop similar changes spontaneously in the course of aging. These models have revealed, using different methods (mitotic figure counting and incorporation of 5-bromo-2'-deoxyuridine into replicating nuclei), a significant increase in the proliferation indices after reserpine stimulation, which can be partially prevented by adrenal denervation [49,52]. Denervation also causes a significant decrease in the proliferation index in nonstimulated animals. This hyperproliferative stage is postulated as a prelude to neoplastic transformation. The chronic persistence of these signals and superimposed abnormalities would lead to the selection of specific cell clones, which becoming dominant, explain the development and progression of neoplasms [50],

Conclusions
Tile currently available evidence on tumorigenesis and clonality in the adren~ medulla can be summarized as follows: 1. Multistep tumorigenesis defines the evolution of pheochromocytomas, as evidenced by the presence of several genetic alteradons. 2. Both the significant association of nonrandom genetic alterations (specially 1 p and 22q interstitial deletions) and the topographic accumulation of genetic deletions at the peripheral tumor corn-partment support a convergent clone selection for these neoplasms. 3. Although many genetic loci show nonrandom abnormalities, the most frequently involved genes locate on chromosome lp, regardless of genetic tumor background (sporadic or inherited predisposition). 4. Most pheochromocytomas should begin as monoclonal proliferations that do not ,always correlate with histopathologic {eatures, particularly in inherited tumor syndromes. 5. Early histopathologic stages, described as adrenal medullary hyperplasias, are defined by hyperproliferative features in animal models and motmclonal patterns in the adrenal nodules from patients with MEN-2a.