| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Editorial |
Genomic Medicine Institute (F.W., C.E.), Cleveland Clinic Lerner Research Institute, and Department of Genetics (C.E.), Case Western Reserve University School of Medicine, Cleveland, Ohio 44195; Department of General Surgery and Transplantation, (F.W), The University of Essen, Essen 45122, Germany; and Cancer Research UK Human Cancer Genetics Research Group (C.E.), University of Cambridge, Cambridge CB2 1XZ, United Kingdom
Address all correspondence and requests for reprints to: Charis Eng, M.D., Ph.D., Cleveland Clinic Genomic Medicine Institute, 9500 Euclid Avenue, Mailcode NE-30 (NE-50 after January 2006), Cleveland, Ohio 44195. E-mail: engc{at}ccf.org.
Medullary thyroid carcinoma (MTC) represents perhaps 10% of all thyroid carcinomas, yet it has captured the imagination of both clinician and investigator. About 25% of all MTC occur as part of the autosomal dominant multiple endocrine neoplasia type 2 (MEN 2) syndrome, which occurs in 1 in 200,000 live births. Caused by germline mutations in the RET protooncogene on 10q11.2, MEN 2 comprises subtypes MEN 2A, MEN 2B, and familial MTC (FMTC), depending on the organs—thyroid, adrenal medulla, and/or parathyroids—involved. With the identification of the disease-causing RET mutations and subsequent genotype-phenotype correlations, i.e. the association of germline mutations at certain codons with specific disease features (age of onset, penetrance of endocrine tumors other than MTC), RET testing in MEN 2 became and continues to serve as the archetype for the practice of genomic medicine (1, 2). RET testing in the setting of MTC has allowed for molecular diagnosis when heritability is in question, for predictive or premorbid testing in individuals who are at risk for MEN 2 but before any features have developed, and for management.
RET testing in MTC/MEN 2 is a success story for genomic medicine but, as time passed, it became obvious that several clinical aspects could not be entirely explained by these traditional mutations in RET. For instance, within some MEN 2 families, the age of onset and severity of disease vary considerably for the affected family members despite their carrying identical germline RET mutations. Furthermore, whereas genotype-phenotype correlations can help predict the probability, i.e. likelihood, of developing specific component tumors, currently it is not possible to predict precisely which individual patient might suffer from parathyroid hyperplasia or pheochromocytoma and at what age they will develop them. One clue to phenotypic modulation, although it was not obvious at that time, was the parable-like observation that certain germline RET mutations originally described in MEN 2A and FMTC, C618R and C620S, also were described in individuals and families segregating both MEN 2A/FMTC and Hirschsprung disease (HSCR) or aganglionosis of the gut (3). Shortly after germline gain-of-function RET mutations were described in MEN 2, germline loss-of-function RET mutations were found in a subset of HSCR (4). The paradoxical observation of a presumptive gain-of-function mutation (C618R, C620S) causing both MEN 2 and HSCR was subsequently explained by observing that RET receptors with missense mutations in the extracellular domain become stuck in the Golgi and endoplasmic reticulum so that the number of mutant receptors on the cell surface is reduced (5, 6).
In this issue of the journal, Baumgartner-Parzer et al. (7) evaluated the phenotype-modifying effect of six single-nucleotide polymorphism (SNP) loci in the RET protooncogene. The group analyzed 3 exonic SNPs (L769L, S836S, and S904S) that do not alter the coding amino acid. Many other SNPs exist within RET, so why did the authors choose these to analyze? For one, the S836S SNP has already been shown to be associated with sporadic MTC with a relative risk of almost 3 (8, 9, 10). Interestingly, in HSCR disease, there was a trend against association with S836S, which lends credence to its association with sporadic MTC (9, 10). Furthermore, synonymous changes might not only cause altered rates of transcription but also, due to changed mRNA structure and stability, translational efficiency may be affected. Another of the SNPs analyzed (G691S) changes the nonpolar amino acid glycine into the polar amino acid serine. Such an alteration can substantially alter protein processing, folding, subcellular localization, or functional properties. Two intronic SNPs have been included in this study. Despite being located in nontranscribed regions, these types of SNPs can have substantial functional effect by altering splice or branch sites or, if located in or near the promoter region, transcription (11).
Easy access to affordable high-throughput sequencing/genotyping techniques has led to a torrent of studies reporting on the association of one or the other genetic variants and disease or disease outcome. One major limitation of many association studies is that they are underpowered. For instance, a sample size over 1000 is required to detect an odds ratio of 1.5 (i.e. a patient with the variant is 1.5 times more likely to have the disease or the outcome) when the suspected susceptibility variant occurs with a frequency of 0.2 (20% of the population). Furthermore, as in this study, often multiple loci and different subgroups are tested to identify an association. Today over 230 SNPs have been reported to the curated databases, of which 55 have been identified as part of the human genome project. Only a limited, and perhaps selected, number of these RET variants are tested for an association. With smaller sample sizes and considering the high random frequency of polymorphisms, how can we differentiate true biologically meaningful associations, spurious associations, and those that occur just by chance? One important way to identify a true disease-modifying alteration is presented by Borrego et al. (12) in a series of studies. In the first study, within an affected family with RET germline mutations for which the phenotype of the members are known, the disease-modifying effect of the A45A polymorphism could be identified (12). This anecdotal finding then served as a springboard to perform subsequent association studies which validated these findings (13, 14). More importantly, these findings were confirmed by a series of independent association studies as well.
Another example in which careful clinical observation led to a good scientific hypothesis involves germline RET V804, normally associated with very-late-onset disease. As described in Ref.15 , Magalhaes and colleagues observed that the only case with early-onset disease in a RET V804M FMTC family had the de novo presence of SNP L769L. Following up on this single family observation, Wiench et al. (15) reported that the L769L polymorphism is more frequent in patients with sporadic MTC and younger than 30 yr old. In a sense, this association is not surprising as the S836S SNP and the L769L SNP appear to be in linkage disequilibrium (S838S and L769L co-occur) in at least some populations (10, 14, 16) (unpublished observation cited in Ref.15). Interestingly, Baumgartner-Parzer et al. (7) showed in their study population, consisting of 22 patients with the FMTC phenotype, the association between the F791Y germline mutation and the L769L polymorphism. These authors did not report any effect on the age of onset of this polymorphism. The high frequency of F791Y within their study population also might suggest either a penetrance-modifying effect and, hence, coming to medical attention, or ascertainment bias. Furthermore, in contrast to previous reports (8, 9, 10, 16), Baumgartner-Parzer et al. did not find linkage disequilibrium (LD) between L769L and S836S. In this respect, we need to be aware that association (co-occurrence of alleles and phenotypes) and linkage (the relation between genetic loci) may be two distinct phenomena. However, the best unifying explanation for the high frequency of F791Y mutations and the lack of LD between L769L and S836S in this population is the existence of a potential founder effect and the possibility that this founding event occurred in a population distinct from those in which S836S is associated with MTC and is in LD with L769L. This is plausible in the context of the MTC and HSCR series from Western Andalucia, Spain (8) or Italy (16) but is more puzzling when considering the German series (Ref.10 and our unpublished data) unless we postulate that this event is relatively recent.
As for most association studies, this study by Baumgartner-Parzer et al. (7) also violates the basic assumption that the control group is randomly picked from the general population so that the former is a true representation of the latter. In fact, commonly, the control group is selected from 1) relatives/friends of the patient, 2) hospital staff, or 3) (student) volunteers. In addition, controls are commonly recruited from the urban, and often ethnically diverse, areas surrounding larger hospitals, which could lead to admixture and/or population stratification (presence of population subgroups). Therefore, cohort studies are in general superior to case-control studies because they more accurately represent the true population. Due to the large sample sizes and careful selection required, this approach is often not practicable. However, the studies by Borrego et al. (13) showed it is feasible to propose that cases can be considered representative of the population if the study is done in a major referral center for a particular region.
The authors addressed the issue of selection bias in the control population with an interesting approach, which may be used as a model for future association analyses, but with some caveats. It is estimated that up to 30% of the general population harbors C cell hyperplasia and about 60% of apparently healthy persons show elevated calcitonin (hCt) levels (16). Baumgartner-Parzer et al. (7) adjusted for this parameter and included two control groups, one with normal and one with elevated hCt. Although no significant differences in the minor allele frequency between these two groups were detected for five of the six loci analyzed, the IVS14-24 SNP frequency differs significantly. In addition, these authors found the IVS14-24 SNP occurred most commonly in patients with sporadic MTC (27.7%) and not in patients with FMTC (11.4%). The IVS14-24 SNP was initially identified in one study by Gath et al. (17) and associated with HSCR. However, this IVS14-24-HSCR association could not be replicated by another group (18), likely due to the high minor allele frequency in the control population. It is interesting to note that the study by Baumgartner-Parzer et al. (7) reports a pronounced difference of the IVS14-24 minor allele frequency between the normal hCt group (6.3%) and the healthy control group with elevated hCt (18.3%, P = 0.016). The latter frequency is similar to that identified in the control group of the negative association study, indicating that the minor allele frequency in the control population can easily be biased. On the other hand, the clever manner of addressing the presence of elevated hCt levels in the normal population may also have been a pitfall of the Baumgartner-Parzer et al. study. If, indeed, normal controls should be a faithful representation of the general population, then perhaps Baumgartner-Parzer et al. should not have used the two separate series of controls but should have mixed the two control series in a ratio of 60% with elevated hCT and 40% without.
We have come full circle now and readers might be even more confused than they were before reading this editorial and may be asking why, given all the detractions, the paper by Baumgartner-Parzer et al. (7) was published at all. The reason, and it is an important one, is because variant-association studies, variant-variant interactions, and variant-environment interactions are here to stay and these studies or their spin-offs may provide data for evidence-based practice of genomic medicine. Are there major limitations to variant-association studies today? The answer is an emphatic yes. How can we address these limitations to best serve our patients? For one, rigorous parameters need to be employed to dissect out the few true RET modifiers among a vast amount of random associations. Such parameters might include that, for example, ancestral markers need to be analyzed to evaluate for population stratification/admixture, and close attention has to be paid to the accurate and precise phenotype classification. It might also be intriguing to propose an approach termed "reverse phenotyping" for RET polymorphisms. Here, the genotype/haplotype is identified in a cohort study and then associated with phenotypical factors, analogous to an unsupervised analysis used in global gene expression analysis that avoids assumptions on the phenotype to identify hidden patterns. Clearly, this is a task beyond the scope of a single group; what would be of help to address this question is a curated database that can hold not only haplotype information but also full clinical data that would allow a consortium to perform meta-analyses, data remining, and validation studies. We are fortunate that, with the International RET Consortium, the infrastructure for such an effort would exist for endocrine neoplasias—let’s make use of it.
Footnotes
RET-related work in our laboratory is supported by Grant 1R01HD39058 from the National Institutes of Health (to C.E.). C.E is a recipient of the Doris Duke Distinguished Clinical Scientist Award.
Abbreviations: FMTC, Familial MTC; hCt, calcitonin; HSCR, Hirschsprung disease; LD, linkage disequilibrium; MEN 2, multiple endocrine neoplasia type 2; MTC, medullary thyroid carcinoma; SNP, single-nucleotide polymorphism.
Received September 8, 2005.
Accepted September 12, 2005.
References
This article has been cited by other articles:
 |
|
 |
|
  A. Cerrato, V. De Falco, and M. Santoro Molecular genetics of medullary thyroid carcinoma: the quest for novel therapeutic targets J. Mol. Endocrinol., October 1, 2009; 43(4): 143 - 155. [Abstract] [Full Text] [PDF]
|
|
Read all eLetters
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
