| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Department of Internal Medicine, Division of Nephrology and Hypertension (B.U.B., M.G., S.G., B.M., H.P.H.N.), and Department of Surgery (G.K.), Albert Ludwigs University of Freiburg, 79106 Freiburg, Germany; Clinical Cancer Genetics and Human Cancer Genetics Programs, Ohio State University Comprehensive Cancer Center (C.E.), Columbus, Ohio 43210; and Clinical Research Center Human Cancer Genetics Research Group, University of Cambridge, Cambridge CB2 2QQ, United Kingdom
Address all correspondence and requests for reprints to: Dr. Bernhard Bender, Medizinische Universitätsklinik, Hugstetterstrasse 55, 79106 Freiburg, Germany. E-mail: bender{at}med1.ukl.uni-freiburg.de
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
As an alternative to structural VHL alterations, there are reports suggesting that hypermethylation as an epigenetic phenomenon might be involved in VHL inactivation. Hypermethylation of a cytosine/guanosine dinucleotide-rich region (CpG island) within VHL has been described in 5 of 26 sporadic renal cell carcinomas, 2 VHL-associated renal cell carcinomas, and 4 VHL hemangioblastomas (9, 16).
Although much work has been performed to examine genetic and epigenetic alterations of VHL in VHL-related renal cell carcinomas and hemangioblastomas, it is surprising that published studies about VHL-associated pheochromocytomas comprise only a maximum of five tumors in the largest series (7, 8, 9, 17, 18, 19). Therefore, we sought to examine a large population-based series of VHL-related pheochromocytomas for somatic intragenic VHL mutations, LOH of markers on chromosome arm 3p, including the VHL locus at 3p25–26, hypermethylation of the upstream VHL-CpG island, and LOH of markers at chromosomal arms with putative relevance for MEN 2 and sporadic pheochromocytomas, namely 1p and 22q (18, 20, 21, 22). Furthermore, all tumors were analyzed for somatic RET protooncogene mutations within exons 10, 11, 13, and 16; as germline mutations of these exons are responsible for most cases of MEN 2 (23), by inference somatic RET alterations might be involved in the development of both sporadic and familial pheochromocytomas (19, 24, 25).
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Fifty-three pheochromocytoma samples were obtained from 38 patients who were consecutively treated over the last 10 yr at the University of Freiburg Medical Center. The University of Freiburg is the major referral center for the Black Forest region, and so patients treated in this center may be considered population based. All 38 patients donated peripheral blood from which germline DNA was extracted (see below). The classification of all tumors into VHL-associated and sporadic pheochromocytomas is based on extensive genetic and clinical examinations. The VHL germline mutations have been previously described (26, 27). All of the patients gave informed consent.
Clinical investigations
All patients presenting with apparently sporadic pheochromocytoma were subjected to thorough history and physical examination to exclude the clinical diagnosis of VHL. Objective studies included direct ophthalmoscopy and magnetic resonance imaging of the brain, spinal cord, and abdomen, as described in detail previously (28).
Tissues and DNA extraction
Germline DNA from peripheral blood leukocytes and somatic DNA from pheochromocytoma were isolated by phenol/chloroform extraction according to standard protocols (29). Because pheochromocytomas are solid tumors, no microdissection was necessary. The portion of the surgical specimen with macroscopic tumor was dissected free of surrounding normal tissue, and the former was used for DNA extraction.
Gene deletion analysis
In preparation for VHL Southern analysis, DNA was digested with EcoRI according to the manufacturer’s guidelines (New England Biolabs, Inc., Beverley, MA), separated on 0.6% agarose gels, and transferred to nylon membranes by standard techniques (29). Southern filters were hybridized with two VHL gene-specific PCR amplicons of 603 bp (primer pair, 5'-CCT CGC CTC CGT TAC AAC AGC CTA-3' and 5'-GAC CGT GCT ATC GTC CCT GC-3') and 643 bp (primer pair, 5'-AGG CAT TGT GAT GTT TAG GGG CA-3' and 5'-CTC GTA TTT TCT GTG CGG ATG G-3') in length. These probes were labeled by incorporation of digoxigenin-11-deoxy-UTP during PCR. VHL gene-specific fragments were detected with antidigoxigenin Fab fragments conjugated to alkaline phosphatase according to the manufacturer (Roche Molecular Biochemicals, Mannheim, Germany).
Mutation analysis
In preparation for single strand conformation polymorphism (SSCP) analysis of the VHL gene, all three exons were prepared by PCR with four primer pairs, previously described, spanning the entire coding and flanking regions of VHL shown previously to contain point mutations (10). The amplicons were electrophoresed through large acrylamide gels (0.5 x MDE acrylamide gel solution, BioWhittaker, Inc., Rockland, ME; in 0.6 x Tris-Boric acid-EDTA buffer; for exon 2, gel included 5% glycerol) at 2 watts for 16 h at room temperature. The bands were visualized by silver staining according to established protocols (30, 31). Aberrant SSCP bands were cut out from the gel, dissolved in water, and reamplified using the same oligonucleotides as in the primary PCR. Reamplified samples were sequenced with nested fluorescence-labeled primers using the dideoxy method and a semiautomatic sequencer (Alf, Pharmacia Biotech, Uppsala, Sweden).
Germline DNA samples from patients without a germline VHL gene mutation were tested by SSCP for mutations within RET exons 10, 11, 13, and 16 according to protocols previously described (32). In addition, all tumors were investigated for somatic RET mutations within these exons.
Methylation analysis
The CpG island within exon 1 of VHL includes the transcriptional start site of the gene. Hypermethylation of this area was examined with a multiplex PCR as described by Prowse and co-workers with two primer pairs after digestion of the template DNA with a methylation-sensitive enzyme (9). One set of primers amplifies exon 3 as a control for the PCR (5'-CTG AGA CCC TAG TCT GCC ACT GAG-3' and 5'-CAA AAG CTG AGA TGA AAC AGT GTA-3'). The other primer set flanks an EheI restriction enzyme site located in a CpG island within exon 1 (5'-GAG GCA GGC GTC GAA GAG TAC GGC-3' and 5'-GAC TGC GAT TGC AGA AGA TGA CCT-3'). If the CpG island is methylated, then EheI will not cleave the recognition site, and PCR across the island would result in a product. If the site is unmethylated, EheI will cleave the site, and no product results after PCR. We took the smallest amount of enzyme that was sufficient for a complete digestion of germline DNA to avoid overdigestion of the tumor samples. Therefore, 1 µg genomic and tumor DNA was digested with 10 U EheI (MBI Fermenta, St. Leon-Rot, Germany) and buffer Y+/Tango in a total volume of 150 µL at 37 C for 16 h. Before PCR, samples were purified with the GeneClean II DNA purification kit (Bio 101, Vista, CA) according to the supplier. The digested DNA was redissolved in 20 µL Tris-ethylendiamine tetraacetate buffer, of which 2 µL were used for the following multiplex PCR with a total volume of 15 µL. The PCR products were separated on 2% agarose gels containing ethidium bromide and visualized by UV light.
LOH analysis
LOH analysis was performed using highly informative microsatellite repeat markers located at 1p21 (AMY2B), 1p21–32 (D1S311), 1p32 (MYCL1), 1p35–36 (D1S160), 3p13–14.1 (D3S1542), 3p14.3 (D3S1514), 3p24.2–25 (D3S1537), 22q11.2 (F8VWFP), 22q12 (D22S268), and 22q13-qter (D22S304) using primer sets according to sequences given (33, 34, 35, 36, 37, 38, 39). In addition, for investigation of the VHL locus at 3p25–26, we used a PCR-generated AccI restriction fragment length polymorphism located in the 3'-untranslated region of VHL (40). After electrophoresis on large acrylamide gels under nondenaturing conditions (MDE acrylamide gel solution, BioWhittaker, Inc.), the gels were silver stained and dried (31). LOH was defined by a reduction of band intensity of more than 50% compared with the remaining allele of the tumor DNA.
Statistical analysis
All calculations were made using Fisher’s exact test, and significance was defined as P < 0.05.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
C mutation
(26). MEN 2 was excluded in the 17 pheochromocytoma
patients without germline VHL mutation. Thus, 36
pheochromocytomas belonging to 21 VHL patients and 17 sporadic
pheochromocytomas belonging to 17 patients without VHL or MEN 2 were
studied for somatic VHL and RET mutations, LOH of
1p, 3p, and 22q loci, and methylation of the VHL CpG
island.
No somatic intragenic RET or VHL mutations were
noted in any of the 53 pheochromocytomas. Hypermethylation of the exon
1 CpG island was not found in any pheochromocytoma of either subtype
(Fig. 1
).
|
and 2).
|
|
). For this part of the study, an
informative tumor was one with at least one informative locus per
chromosome arm; LOH at an arm was scored when at least one marker in
that chromosome arm showed LOH. LOH at any 3p locus was found to occur
in 32 of 35 (91%) informative VHL pheochromocytomas, whereas only 4 of
17 (24%) informative sporadic tumors showed an allelic deletion at
chromosome 3p. The difference in frequency of 3p LOH between VHL and
sporadic pheochromocytomas is statistically significant (P
<< 0.0001). In contrast to the 3p data, LOH at 1p occurred in only 5
of 33 (15%) informative VHL pheochromocytomas compared with 12 of 17
(71%) informative sporadic tumors (P = 0.0003).
Differential LOH frequencies within 22q was also significant, but not
to such an extent as those for 1p and 3p; 6 of 29 (21%) informative
VHL tumors and 8 of 15 (53%) informative sporadic pheochromocytomas
demonstrated LOH at 22q (P < 0.05).
|
C germline
mutation and those without this mutation (P > 0.05).
The frequency of 1p LOH among tumors with the germline c505TC
mutation might be lower (2 of 24 [8.3%] informative tumors) than
those without this mutation (3 of 9 [33%] informative tumors;
P = 0.1). This difference might become statistically
significant given larger numbers. In contrast, only tumors with the
c505TC mutation demonstrated a 22q LOH; however, this was also not
significant (P = 0.15).
Finally, we correlated LOH of markers on each examined chromosome arm
with age at diagnosis, tumor location (adrenal and extraadrenal), and
benign or malignant status. Interestingly, all four malignant
pheochromocytomas (three sporadic and one VHL), all had LOH at 1p; none
of the three sporadic malignant tumors had 3p LOH (Table 2
). These
differences did not reach statistical significance because of the small
number of malignant tumors.
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Because of a founder effect in the region of Germany for which the
University of Freiburg is the major referral center, the great majority
of investigated VHL cases carry the c505TC
(Tyr98His) germline mutation. The remainder of
VHL germline mutations are also missense, which is not
surprising because genotype-phenotype analyses have revealed that
missense mutations are associated with development of pheochromocytoma
in VHL families (41). Of note, the VHL patients with two
other recurrent germline mutations c746TA and c775CG are,
respectively, related too. Because of founder effects that account for
several recurring germline VHL mutations in our population,
to our knowledge this should not influence the pattern of subsequent
LOH, which are somatic events. After all, 3p LOH has been described
before in VHL-related pheochromocytomas originating from nonfounder
populations, albeit in very small series (7, 8, 9, 18).
The genetic alterations, i.e. specific gene mutations, remain elusive in sporadic pheochromocytoma. Although pheochromocytoma is a component tumor of both MEN 2 and VHL, somatic RET mutations occur in less than 10% of sporadic pheochromocytomas and no VHL tumors (19, 24, 25, 42, 43, 44). Somatic VHL mutations have been found in less than 2% of sporadic pheochromocytomas and have not been described as the second genetic hit in VHL-related pheochromocytomas (9, 10, 19, 24). In the present study, therefore, the complete lack of somatic RET mutations in our tumors is not surprising. Further, the occult detection of germline RET mutations in apparently sporadic pheochromocytoma presentations is negligible (<2%) (19). In reality, there would have been no occult germline RET mutations in apparently sporadic presentations of pheochromocytoma. In this particular instance (19), neither the general practitioner nor the endocrinologist recognized the large neck mass in the proband and the proband’s father at the time of pheochromocytoma presentation. Our present study also demonstrates that, in contrast to renal cell carcinomas (9, 16), epigenetic silencing of the VHL gene in both sporadic and VHL-related pheochromocytomas by hypermethylation of the upstream CpG island is not a prominent mechanism.
Pheochromocytomas from VHL germline mutation-positive cases are believed to abide by the Knudson two-hit mechanism of tumorigenesis. To date, this was viewed as dogma based on 4 studies encompassing a total of only 12 VHL pheochromocytomas and a broad range of 3p LOH frequencies with an average of 50% (7, 8, 9, 18). Our study, which is based on 36 VHL-associated pheochromocytomas, has demonstrated for the first time that indeed a complete and biallelic 2-hit VHL inactivation is a necessary pathogenetic mechanism for these tumors. In more than 91% of such tumors, the second hit is somatic deletion of the 3p region encompassing the remaining wild-type VHL allele. The lack of LOH at the VHL region in 2 informative VHL tumors (T4 and T23) may be explained by heavier contamination of the tumors by nonneoplastic cells. If so, then virtually all VHL-related pheochromocytomas have 2 structural hits at the VHL locus. If, however, we speculate that these tumors truly have no LOH of 3p locus, then perhaps LOH at other loci contribute to tumorigenesis (e.g. tumor 23). What might also be highlighted by our data are that other genetic alterations, affecting tumor suppressor genes on 1p and 22q, might be necessary for the progression of pheochromocytoma. In contrast to VHL tumors, sporadic pheochromocytomas have a relatively low frequency of 3p LOH, which supports 4 other reports that demonstrated LOH at 3p in 35% of a total of 52 sporadic tumors (8, 18, 21, 22). These studies together with the observation of a lack of somatic intragenic VHL mutations in sporadic tumors gave the first hint that VHL-related and sporadic pheochromocytomas do not share common pathogenic pathways. Our study for the first time conclusively demonstrates that although the VHL gene plays a principal role in the tumorigenesis of VHL pheochromocytoma, sporadic pheochromocytomas only use the VHL genetic pathway in the minority of cases. Instead, an as yet undiscovered gene(s) residing on 1p plays a major role in sporadic tumorigenesis. Tumor suppressor genes on 22q also contribute to sporadic pheochromocytoma genesis in no small way.
Our observations of genetic alterations in VHL-related and sporadic pheochromocytoma are in contrast to observations that have been made among inherited cancer syndromes with involvement of tumor suppressor genes to date; sporadic counterparts of component tumors in several inherited cancer syndromes share a common genetic etiology and pathogenic pathway. The classic example is retinoblastoma, on which the Knudson two-hit hypothesis is based (6). In heritable retinoblastoma, the first hit is germline mutation of the RB1 gene, and the second hit is a somatic mutation or deletion of RB1 (45, 46). In sporadic retinoblastoma, two somatic mutations affecting RB1 are the genetic basis of tumorigenesis (47). This is also true of TP53 mutations in Li-Fraumeni syndrome and its sporadic counterpart tumors osteosarcoma, soft tissue sarcomas, and glioblastomas (48, 49), and APC mutations in familial adenomatous polyposis and a subset of sporadic colon cancer (50, 51). With respect to the VHL gene, its biallelic inactivation is a well known feature in VHL-related as well as sporadic renal clear cell carcinomas and hemangioblastomas (10, 11, 12, 14, 15). However, there are other examples in the literature where the genetic pathway in a hereditary cancer is different from the sporadic counterpart. For example, although germline BRCA1 and BRCA2 mutations are associated with the hereditary breast-ovarian cancer syndrome, less than 1% of sporadic breast cancers or ovarian cancers harbor somatic BRCA1 or BRCA2 mutations (52, 53). Further, breast carcinomas from BRCA1 mutation carriers have been described to harbor different genetic alterations compared with their sporadic counterparts (54, 55, 56). Thus, the different frequencies of VHL inactivation, and LOH at 1p, 3p, and 22q between VHL pheochromocytoma and sporadic pheochromocytoma suggest that in contrast to renal clear cell carcinomas and hemangioblastomas, different genetic pathways might be involved in hereditary vs. sporadic disease, a situation more akin to the BRCA loci and breast cancer (above).
The similar LOH distribution among the investigated markers in some VHL patients with multiple pheochromocytomas is an interesting observation. For example, patients P15, P18, and P20 have had multiple tumors, with an intraindividual identical LOH pattern of all investigated chromosomal markers. One possible explanation is a monoclonal origin of these tumors, and this might be true for patient 15, in whom the multiple tumors have been located within one adrenal. On the other hand, tumors T30 and T31 from patient 18 were bilateral, and in patient 20, one pheochromocytoma was in the adrenal, and the other was extraadrenal. There are three possible explanations. First, these two pairs of multiple tumors might be not be separate primaries, but one may be a metastasis of the other. Second, perhaps certain specific germline VHL mutations are associated with specific subsequent somatic genetic alterations. This has been shown in colon cancers from individuals with germline APC mutations (57). Third, chance occurrence of similar LOH patterns cannot be excluded. Because several VHL individuals have multiple tumors, some with identical LOH patterns, we also recompared LOH frequencies at the three loci between sporadic and VHL-related pheochromocytoma, but counting multifocal tumors within a single adrenal in a single patient as one tumor. When these were reanalyzed in this manner, the differences were still found to be statistically significant.
In addition to demonstrating the high frequency of 1p involvement in sporadic pheochromocytoma genesis, our data might also suggest that malignant pheochromocytomas in particular might be associated with LOH of 1p. However, the number of malignant tumors in our series is small, and 1p LOH is common among all sporadic pheochromocytomas. Nonetheless, the association of 1p alteration and malignant disease might eventually be conclusively proven given previous observations describing allelic loss of chromosome 1p as a predictor of an unfavorable outcome in patients with neuroblastoma, another tumor with neural crest derivation like pheochromocytomas (58, 59). With regard to the precise regions of 1p that harbor putative tumor suppressors, there might be at least 2 regions of interest; 1 area seems to be centromeric to 1p32; the other is possibly located telomeric to this subband. These data correspond to those of Schleiermacher and co-workers (60), who proposed 1 centromeric and 1 telomeric tumor suppressor gene locus on chromosome 1p after performing fine structure LOH analysis with 25 polymorphic markers across 1p in 60 neuroblastomas. In addition, there are several reports of a variety of tumors of neuroectodermal origin suggesting the presence of multiple tumor suppressor loci on chromosome arm 1p (61, 62, 63, 64). To date, however, no gene at this chromosomal arm with frequent biallelic inactivation in pheochromocytomas (and other tumors with 1p deletions) has been found. In particular, investigation of the recently identified p73 gene on 1p36.2–3 was unrevealing in this regard; with the exception of only 1 breast cancer and 1 neuroblastoma, no somatic alterations have been detected in any tumor of several tissues (65, 66) (Bender, B. U., et al., unpublished data). However, it is interesting to note that the elongin A gene is located on 1p36.1. Elongin A forms a complex with elongin B and elongin C. The VHL protein competes with elongin A in binding the binary elongin B/C complex; the elongin A/B/C complex stimulates ribonucleic acid transcription of several genes, whereas the pVHL/B/C complex is inactive (67). Given this functional relationship between pVHL and the elongins, it is unclear how loss of elongin A leads to tumorigenesis.
In conclusion, our data suggest that VHL gene inactivation is an event essential for VHL-related pheochromocytoma genesis. Additional, as yet undefined steps might be required as well. In contrast, the pathogenesis of sporadic pheochromocytomas does not rely on VHL gene inactivation. Instead, a gene(s) on chromosome arm 1p and, to a lesser degree, 22q might play a prominent role in these nonfamilial tumors. Further studies are necessary to identify the responsible tumor suppressor genes on these chromosomes.
|
Footnotes |
|---|
Received January 26, 2000.
Revised June 14, 2000.
Revised August 15, 2000.
Accepted August 23, 2000.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C D E Margetts, D Astuti, D C Gentle, W N Cooper, A Cascon, D Catchpoole, M Robledo, H P H Neumann, F Latif, and E R Maher Epigenetic analysis of HIC1, CASP8, FLIP, TSP1, DCR1, DCR2, DR4, DR5, KvDMR1, H19 and preferential 11p15.5 maternal-allele loss in von Hippel-Lindau and sporadic phaeochromocytomas Endocr. Relat. Cancer, March 1, 2005; 12(1): 161 - 172. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Y. Kim and W. G. Kaelin Role of VHL Gene Mutation in Human Cancer J. Clin. Oncol., December 15, 2004; 22(24): 4991 - 5004. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Santoro, R. M. Melillo, F. Carlomagno, G. Vecchio, and A. Fusco Minireview: RET: Normal and Abnormal Functions Endocrinology, December 1, 2004; 145(12): 5448 - 5451. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Cascon, S Ruiz-Llorente, M F Fraga, R Leton, D Telleria, J Sastre, J Jose Diez, G Martinez Diaz-Guerra, J A Diaz Perez, J Benitez, et al. Genetic and epigenetic profile of sporadic pheochromocytomas J. Med. Genet., March 1, 2004; 41(3): e30 - 30. [Full Text] [PDF]
|
|
|
|
 |
|
 C. A. Koch, K. Pacak, and G. P. Chrousos The Molecular Pathogenesis of Hereditary and Sporadic Adrenocortical and Adrenomedullary Tumors J. Clin. Endocrinol. Metab., December 1, 2002; 87(12): 5367 - 5384. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A Cascon, A Cebrian, S Ruiz-Llorente, D Telleria, J Benitez, and M Robledo SDHB mutation analysis in familial and sporadic phaeochromocytoma identifies a novel mutation J. Med. Genet., October 1, 2002; 39(10): e64 - 64. [Full Text] [PDF]
|
|
|
|
 |
|
 C. A. KOCH, A. O. VORTMEYER, Z. ZHUANG, F. M. BROUWERS, and K. PACAK New Insights into the Genetics of Familial Chromaffin Cell Tumors Ann. N.Y. Acad. Sci., September 1, 2002; 970(1): 11 - 28. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Young, B. E. Baysal, A. Deb, and W. F. Young Jr. Familial Malignant Catecholamine-Secreting Paraganglioma with Prolonged Survival Associated with Mutation in the Succinate Dehydrogenase B Gene J. Clin. Endocrinol. Metab., September 1, 2002; 87(9): 4101 - 4105. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Baghai, G. B. Thompson, W. F. Young Jr, C. S. Grant, V. V. Michels, and J. A. van Heerden Pheochromocytomas and Paragangliomas in von Hippel-Lindau Disease: A Role for Laparoscopic and Cortical-Sparing Surgery Arch Surg, June 1, 2002; 137(6): 682 - 689. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. C. T. Aguiar, G. Cox, S. L. Pomeroy, and P. L. M. Dahia Analysis of the SDHD Gene, the Susceptibility Gene for Familial Paraganglioma Syndrome (PGL1), in Pheochromocytomas J. Clin. Endocrinol. Metab., June 1, 2001; 86(6): 2890 - 2894. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Hoffman, M. Ohh, H. Yang, J. M. Klco, M. Ivan, and W. G. Kaelin Jr von Hippel-Lindau protein mutants linked to type 2C VHL disease preserve the ability to downregulate HIF Hum. Mol. Genet., May 1, 2001; 10(10): 1019 - 1027. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab.< |
