| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
Departments of Medicine (X.W., B.M., I.D.H.), Biochemistry and Molecular Biology (N.L.E.), and Laboratory Medicine and Pathology (J.R.G., S.K.G.G.), Mayo Clinic and Foundation, Rochester, Minnesota 55905; and Department of Pathology and Molecular Medicine, Wellington School of Medicine (K.F., B.D., S.K.G.G.), Wellington, New Zealand
Address all correspondence and requests for reprints to: Dr. Stefan K. G. Grebe, Department of Laboratory Medicine and Pathology, Mayo Clinic and Foundation, 200 1st Street SW, Rochester, Minnesota 55905. E-mail: grebs{at}mayo.edu.
Abstract
There is strong evidence in many tumor types, including thyroid cancer, for a novel tumor suppressor gene (TSG) at 17p13. To identify the putative thyroid 17p13 TSG we mapped thyroid tumor loss of heterozygosity (LOH) at high resolution within this region. We examined 20 typical follicular thyroid carcinomas (FTC), 19 Hurthle cell carcinomas (HCC), 15 papillary thyroid carcinomas (PTC), and 7 follicular adenomas (FA) for LOH at 17p13 using 18 probes. Complete clinical follow-up data were available for all patients. We confirmed a high 17p13 LOH rate in FTC (18 of 20) and HCC (13 of 19) and showed an association between 17p13 LOH and advanced tumor grade. Only 4 of 15 PTC and 1 of 7 FA displayed 17p13 LOH. In the HCC we identified a narrow minimal common deleted region between D17S1308 (285 kb from the p-telomer) and D17S695 (696 kb from the p-telomer). This region was flanked centromerically by a breakpoint cluster, further suggesting nonrandom deletion. All but 1 of the PTC and FA with 17p LOH and 50% of the affected FTC also showed LOH in this region.
These data suggest that a TSG, involved in HCC pathogenesis, is contained within the D17S1308-D17S695 interval. There are several potential candidate TSGs in this region that are worthy of further study.
DURING THE LAST decade our understanding of the initiating events in thyroid tumorigenesis has increased considerably. It is now widely accepted that the most common thyroid tumor morphotype, papillary thyroid carcinoma (PTC), probably arises de novo from normal follicular epithelial cells (1). The initiating somatic genetic events in this transformation often involve genetic rearrangements, which lead to constitutive activation of tyrosine kinase receptors, most commonly RET or NTRK (1). By contrast, follicular thyroid carcinoma (FTC) seems do develop in a multistep fashion from initially benign hyperplastic nodules through follicular adenomas (FA) into malignant FTC (1, 2). There is increasing genomic disarray during each step of this process, with activation of the RAS oncogenic pathway probably representing an important early event in FTC tumorigenesis (1, 2, 3, 4, 5).
Both PTC and FTC may subsequently acquire further somatic genetic changes, which can result in tumor dedifferentiation and clinical progression. For example, inactivation of the TP53 tumor suppressor gene (TSG) seems to be involved in anaplastic transformation (6, 7). However, only a small proportion of PTC or FTC advance to this stage. Many more display lesser degrees of tumor progression, which nonetheless can be associated with adverse clinical outcomes. Little is known about the somatic genetic changes that underlie these intermediate degrees of thyroid tumor dedifferentiation. On a relatively coarse genetic level there is an overall increase in tumor aneuploidy, particularly in FTC (8). Certain chromosomal regions seem to be preferentially involved, suggesting that they may harbor TSGs (9, 10, 11, 12, 13, 14, 15). One such region is the telomeric portion of the short arm of chromosome 17. We have previously shown that this region exhibits extremely high rates of allelic imbalance in advanced FTC, and that the known 17p TSG TP53 is not the target of these somatic genetic changes (11). In addition, in our initial cohort of patients, 17p loss of heterozygosity (LOH) was associated with tumor death, suggesting that the unknown molecular target of the genetic changes was of prognostic and biological relevance. These findings were consistent with observations in other human tumor types, such as breast cancer and medulloblastoma, where 17p LOH without TP53 inactivation had also been observed and found to correlate with adverse clinical outcomes (16, 17, 18, 19, 20). In conjunction with our own findings these data suggest that 17p13 may harbor a novel oncogene or TSG that plays a role in thyroid carcinoma progression.
To identify this putative thyroid tumor gene we decided to map the region of allelic imbalance at 17p13 more precisely. We therefore extended our previous studies to a larger cohort of clinically and histologically well characterized tumors, mainly typical FTC and oxyphilic FTC [Hurthle cell carcinoma (HCC)], using a series of well mapped and closely spaced microsatellite markers.
Materials and Methods
Patients and tissue specimens
All studies were approved by the Mayo Clinic institutional review board and the Wellington ethics committee.
We studied tumor specimens from patients who had undergone surgery for primary or recurrent thyroid carcinoma at the Mayo Clinic between 1972 and 1999. From this period we selected 20 typical FTC, 19 HCC, 15 PTC, and 7 FA specimens from patients for whom complete follow-up data and sufficient archival tissues were available. For all patients, age at diagnosis, gender, relevant past and family history, and clinical outcomes were recorded up to January 2001. All tumor specimens were staged by the international tumor, node, metastasis (TNM) system and graded according to Mayo Clinic thyroid tumor histological grading protocols. The main demographic and histopathological data of our study subjects and their tumors are summarized in Table 1
.
|
We sectioned formalin-fixed, paraffin-embedded tissue blocks at 10-µm thickness. We stained one slide of each block with hematoxylin and eosin, and this slide was reviewed by an expert endocrine pathologist (J.R.G.), who confirmed the diagnosis and histological grading. Using the stained slides as templates, another anatomical pathologist (B.D.) then microdissected between one and three of the corresponding unstained slides into neoplastic compartments, containing at least 80% tumor cells (typically >90%) and nonneoplastic compartments, containing thyroid tissue from the same slide, which appeared microscopically free of tumor.
After microdissection, we deparaffinized the samples by successive xylene and ethanol washes and incubated them in 150 µl DNA extraction buffer each [2.7 µg/µl proteinase K (Roche, Auckland, New Zealand), 100 mM Tris, and 2 mM EDTA, pH 8] at 55 C for 48 h. We added additional proteinase K (2.7 µg/µl) after 12 h. After heat inactivation of proteinase K, aliquots of the digests were used directly for PCR.
Microsatellite marker selection and PCR
We performed LOH analysis by paired normal-tumor microsatellite PCR, using 18 well mapped microsatellite markers (Table 2). These markers cover the region from approximately 180 kb to about 12,500 kb from the chromosome 17 p-telomer. Figure 1 shows the chromosomal positions of the selected markers, ordered on the basis of the sequence tagged site database (http://www.ncbi.nlm.nih.gov), with supplementary mapping information, if necessary, provided through the Cooperative Human Linkage Center database (http://www.chlc.org), the Genome Database (http://www.gdb.org), the Genetic Location Database (http://cedar.genetics.soton.ac.uk/public_html), and other published chromosome 17 genetic maps (21). There were no conflicts with regard to mapping order or genomic distances of our markers between the different databases. However, estimates for physical distances to the chromosome 17 p-telomer differed for several markers. In these cases we used the distances provided through the human genome project via the STS database as the most likely correct physical distances.
|
|
). Analysis of microsatellite PCR
After PCR, we size-separated the PCR-products on a 377 automated sequencer (PE Applied Biosystems) and analyzed the fluorescent gel data with the GeneScan software package (PE Applied Biosystems). For each informative tumor-control pair of reactions (two alleles visible in control samples), we calculated an allelic imbalance ratio: control-allele 1:control-allele 2/tumor allele 1:tumor allele 2. Because LOH analysis of DNA derived from archival samples can sometimes result in spurious allele ratios, we applied strict allelic imbalance criteria for LOH scoring. Whereas allelic imbalance ratios less than 0.74 or greater than 1.35 are often considered indicative of LOH (22, 23), we only regarded an allelic imbalance ratio less than 0.55 or more than 1.82, corresponding to allelic loss in at least 65% of tumors cells in an 80% pure tumor sample, as definitive LOH. Allelic imbalance ratios of 0.55–0.65 or 1.54–1.82, corresponding to between 45–65% of tumor cells having suffered allele loss, were defined as borderline LOH. We repeated all experiments for samples with borderline LOH scores at least three times and only scored those samples as LOH for which all repeat experiments confirmed at least a borderline LOH classification. For each morphotype we determined a minimal common region of loss/LOH (MCRL) for the tumors with 17p13 LOH. An MCRL corresponds to the smallest region of LOH shared by all tumors with LOH. Because one allele of a TSG is commonly inactivated by genetic deletion, TSGs can be localized to certain genomic regions by determining MCRLs.
We defined breakpoints as regions of LOH immediately adjacent to regions of retained heterozygosity. We considered a marker to display microsatellite instability (MSI) in a given tumor if additional alleles occurred in the tumor sample or the allele size(s) in the tumor was changed compared with the matched nontumorous control sample. Figure 2 shows some examples of noninformative markers, informative markers with and without LOH, and MSI.
|
Continuous demographic, clinical, and histopathological variables were compared using ANOVA with Fisher’s protected least significant difference (PLSD) post hoc t tests for subgroup analysis if appropriate. We compared stage and grade distribution between morphotypes using 
2 tests with appropriate degrees of freedom.
For each tumor, we noted whether any LOH or MSI at 17p13 had occurred and for tumors with LOH we calculated the fractional allele loss rate (proportion of informative alleles lost). We also calculated the LOH rate per marker (number of tumors with LOH at the marker/total number of tumors informative for this marker). We compared the overall LOH rates, fractional allele loss rates, and marker LOH rates between morphotypes, stages, and grades using 2 tests with appropriate degrees of freedom.
We compared Kaplan-Meier survival functions for tumor survival between the different morphotypes, stages, and grades, with and without stratification for 17p LOH, using the log-rank test for significance testing. Cox modeling was used for multivariate analysis.
Results
Patients and tumors
The average patient age at diagnosis was 51.4 yr (range, 13.2–80.9), and the mean follow-up period was 10.8 yr (range, 0.06–32.67). Twelve patients died of their tumors during the follow-up period.
There were few, if any, differences among the four thyroid tumor morphotypes with regard to average patient age at diagnosis, average length of follow-up, and gender distribution. The exceptions were that HCC patients, the group with the greatest average age of 58.1 yr, were significantly older at diagnosis than those with PTC, the youngest group with an average age of 45.1 yr (by post hoc Fisher’s PLSD, P = 0.033), and that mean follow-up in FA (shortest mean follow-up, 4.8 yr) was significantly shorter (by post hoc Fisher’s PLSD, P = 0.02) than in FTC (longest mean follow-up, 12.9 yr). Table 1 summarizes the relevant patient demographic data and tumor histopathological characteristics.
Microsatellite LOH analysis
Microsatellite PCR was informative in an average of 70.3% (range, 47.5–97.1%) of reaction pairs (Table 2). We observed allelic imbalances at 17p in all 4 tumor morphotypes. Tumors fell into 3 broad categories of LOH patterns: those with no LOH or LOH at a single locus (29 of 61), those with fractional allele loss rates between 10–60% (24 of 61), and a small group (all HCC or FTC) of 8 specimens with fractional allele loss rates of 70% or more. There were significant differences between morphotypes in the proportion of specimens with LOH (by 2 test: 3 df, 20.914; P < 0.0001), with fewer FA (1 of 7) or PTC (4 of 15) specimens affected than FTC (18 of 20) or HCC (13 of 19) specimens. In addition, among those specimens with LOH, the extent of genetic imbalance was much larger in FTC and HCC, which had lost an average of 40% (FTC) and 54% (HCC) of informative markers, compared with FA and PTC, with fractional allele losses of only 17% and 7%, respectively (P = 0.018).
Figure 3 maps the distribution of allelic imbalances in the different morphotypes. In the FTC, the allelic imbalance patterns differed significantly between individual tumors with a large minimal common region of loss stretching from D17S1308 (inclusive) at approximately 285 kb from the p-telomer to p53-penta (inclusive) at about 8651 kb from the p-telomer. By contrast, HCC allelic imbalance patterns showed less variability with all 13 tumors with LOH displaying at least one loss within a narrow MCRL, spanning 411 kb, between D17S1308 (inclusive) at 285 kb from the p-telomer and D17S695 (inclusive) at 696 kb from the p-telomer. Of the 4 PTC and the single FA with 17p LOH, all but 1 PTC also had suffered LOH within this HCC-MCRL, but only half of the 18 FTC with 17p LOH had lost markers in the HCC-MCRL.
|
In the FTCs, breakpoints were evenly distributed. However, in the HCCs 6 breakpoints mapped to the D17S695-ABR interval, twice as many than to any of the other 16 marker intervals, which displayed an average of 1.2 breakpoints each. There were too few breakpoints in the FAs and PTCs for meaningful analysis.
No tumor fulfilled the consensus criterion of MSI at more than 35% of tested loci, which is required for diagnosis of high grade MSI (24). A small number of specimens could be classified as low grade MSI: MSI occurred at a single locus in 1 FA specimen, 3 FTC specimens, 2 HCC specimens, and 1 PTC specimens. One additional FTC and HCC specimen each displayed MSI at 2 markers, and a single HCC specimen showed evidence of MSI at 4 of 18 loci.
Clinicopathological correlations
Among the three malignant tumor morphotypes, stage and grade distribution, number of tumor deaths, and median survival time were not significantly different (Table 1).
Across all morphotypes, stage and grade were significant and independent predictors of tumor death (P = 0.0035).
As described above, LOH at 17p was significantly correlated with FTC or HCC morphotype (by 2: 3 df, 20.914; P < 0.0001) and, for the malignant tumors, with higher tumor grade regardless of morphotype (by 2: 2 df, 6.02; P < 0.05). There was no significant relationship between tumor survival and 17p LOH status or 17p fractional allele loss.
The results for LOH involving HCC-MCRL mirrored the overall 17p LOH results.
Discussion
Our study has confirmed several previous observations (10, 11, 12, 13), showing both a significantly higher degree of allelic imbalances in FTC and HCC than in PTC, even when matched for grade, stage, survival, and host characteristics, and a low rate of MSI in all thyroid tumor morphotypes (25). We also confirmed a high 17p allelic imbalance rate in FTC and HCC. This is an observation not universally shared by others (12, 13) and is most likely due to the choice and number of markers studied. The high marker density used in our study may also account for the relatively similar LOH rates in FTC and HCC. In high density LOH mapping, the probability of finding any LOH increases. FTC display enough genomic instability for this increased LOH detection likelihood to result in LOH rates similar to those in HCC. The higher fractional allele loss rates in HCC suggest that HCC nonetheless have a greater degree of genomic instability than FTC, as suggested by the literature (14). We were also able to demonstrate that despite high LOH rates in both FTC and HCC, the 17p13 LOH patterns in these two morphotypes are distinct. This suggests unique targeting, consistent with TSG inactivation, of a narrow genomic region in HCC, but not in FTC. We mapped this region of targeted 17p13 LOH in HCC to a very small MCRL, thus narrowing the position of a putative 17p13 thyroid cancer tumor suppressor gene to about 411 kb. All 13 HCC with 17p LOH had lost at least 1 marker within this narrow genomic segment, and the marker interval flanking the MCRL in the centromeric direction, D17S695 to ABR, contained twice as many breakpoints as any other marker interval. This suggests that the HCC-MCRL represents a genetic region distal to a chromosomal fragile site, a situation predisposing to LOH. The fact that the resultant telomeric LOH affects all HCC specimens with 17p LOH indicates that LOH in this region may provide HCC cells with a selective growth advantage. Further support for this idea comes from the fact that 17p LOH was associated with increased tumor grade, a histopathological feature usually associated with more advanced, more rapidly growing tumors. Most FA and PTC with 17p LOH and 50% of FTC with 17p LOH also showed genetic losses in the HCC-MCRL, and the association of high tumor grade with 17p LOH was shared by these morphotypes. However, few FA and PTC showed 17p LOH, and half of the FTC with 17p LOH showed no LOH in the HCC-MCRL. LOH sites and breakpoints in FTC were also more evenly distributed across the examined region, suggesting a lack of specific targeting of the HCC-MCRL in the typical FTC. It is therefore likely that the putative tumor suppressor gene in the HCC-MCRL is specific for HCC.
Failure to confirm our previous observations of an association between death from thyroid cancer and 17p LOH (11) does not argue against the possibility that chromosome band 17p13.3 harbors a novel TSG. Our previous cohort of patients had significantly more aggressive tumors, resulting in relatively higher event numbers. In addition, by using a much higher density of 17p markers in the current study, we detected 17p LOH in almost all HCC and FTC cases. As a consequence of these two factors, our current sample size was too small to confirm an impact of 17p LOH on tumor survival even though we studied a larger number of tumors than in the previous study.
Because the 17p13 region telomeric to TP53 also suffers frequent LOH in a variety of other tumor types, including breast and ovarian cancer, medulloblastoma, and hepatocellular carcinoma, considerable efforts have been directed toward identifying possible TSGs in this region (16, 17, 18, 19, 20, 26, 27, 28, 29). The two main candidates to date, HIC-1 and DPH2L (also known as OVCA1) (19, 30, 31, 32, 33), map centromeric to our HCC-MCRL and are therefore unlikely to represent the putative HCC TSG. As of February 2002 the NCBI gene sequence maps list 4 confirmed genes (GEMIN4, CGI-150, NXN, and TIMM22) and 17 potential genes, for a total of 21 possible genes, within our HCC-MCRL. The actual number of genes in this region may be slightly smaller or larger than this number. On the one hand, some potential genes may be pseudogenes or portions of larger genes; on the other hand, neither sequencing nor contig ordering has been entirely completed for this genomic region, raising the possibility it could contain a few additional genes. Among the confirmed genes in this region little is known about CGI-150 other than that its protein product shows some homology to the glyoxalase/bleomycin resistance protein/dioxygenase superfamily of proteins. However, GEMIN4 and NXN are potential candidate TSGs/oncogenes. GEMIN4 codes for a DEAD box cofactor involved in RNA splicing and processing (34, 35). There is increasing evidence for a role of disordered RNA processing in human carcinogenesis. We have shown that aberrant splicing of FHIT, TSG101, and TP53 occurs frequently in thyroid neoplasms, particularly in HCC and dedifferentiated FTC (36), whereas others have made similar observations in different human tumor types (37). NXN encodes nucleoredoxin 1, a thiol reductase involved in the regulation of the redux state of transcription factors (38). TIMM22 is also of potential interest, as its protein product forms part of the translocase of the inner mitochondrial membrane protein complex, which controls translocation and targeting of different proteins to the inner mitochondrial compartment (39). This is of particular interest, because HCC cytoplasm is densely packed with mitochondria, giving rise to the eosinophilic staining characteristics of these tumors (40) and suggesting some form of mitochondrial dysfunction. Finally, a fifth gene, HCCS1, has recently been confirmed among the 17 potential genes within the HCC-MCRL. This gene was cloned from a region of frequent LOH in hepatocellular carcinoma, nearly identical in chromosomal position and size to our thyroid HCC-MCRL (41). Sequence comparison of HCCS1 with the various potential gene entries in the NCBI gene sequence maps shows that HCCS1 consists mainly of 3 Unified gene database (UniGene) clusters that were previously thought to code for 3 separate potential genes (XM_008540, XM_091682, and XM_102804). The gene consists of 18 exons distributed over a region of about 230 kb between WI-14673 (inclusive, 
150 kb from 17p-telomer) and D17S926 (inclusive, 383 kb from 17p-telomer) (41). The protein product has no homology with genes of known function, but is a probably a TSG candidate, having shown frequent mutations in hepatocellular carcinoma, under-expression in hepatocellular carcinoma tissue vs. normal neighboring hepatic tissue, and growth-suppressive abilities in cell culture and nude mouse models (41). Moreover, like TIMM22, the HCCS1 protein localizes to mitochondrial structures, making this gene particularly interesting in the context of a possible thyroid HCC tumor suppressor (41).
Once the human genome project sequencing efforts of 17p near completion, some of the remaining 14 potential genes in the thyroid HCC-MCRL plus possibly some additional not yet discovered genes may also become candidate tumor suppressors. At that stage, mutation screening and in situ protein expression profiling should be applied to all serious candidates to finally identify the putative HCC 17p TSG mapped in this study.
Acknowledgments
Footnotes
This work was supported by grants from the University of Otago, the Wellington Division of the Cancer Society of New Zealand, New Zealand Lotteries Health Research (all to S.K.G.G.), NIH Grant CA-80117 (to N.L.E), and funds from the Mayo Foundation and Clinic.
Abbreviations: FA, Follicular adenoma; FTC, follicular thyroid carcinoma; HCC, Hurthle cell carcinoma; LOH, loss of heterozygosity; MCRL, minimal common region of loss; MSI, microsatellite instability; PLSD, protected least significant difference; PTC, papillary thyroid carcinoma; TSG, tumor suppressor gene.
Received May 6, 2002.
Accepted July 19, 2002.
References
B, AP-1, and CREB activation in HEK293 cells. Biochem Biophys Res Commun 274:177–182[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  E. M Pinto, A. E. C. Billerbeck, M. C. B. V. Fragoso, B. B. Mendonca, and A. C. Latronico Deletion Mapping of Chromosome 17 in Benign and Malignant Adrenocortical Tumors Associated with the Arg337His Mutation of the p53 Tumor Suppressor Protein J. Clin. Endocrinol. Metab., May 1, 2005; 90(5): 2976 - 2981. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. M. Reis, E. P.B. Ojopi, F. L. Alberto, P. Rahal, F. Tsukumo, U. M. Mancini, G. S. Guimaraes, G. M.A. Thompson, C. Camacho, E. Miracca, et al. Large-scale Transcriptome Analyses Reveal New Genetic Marker Candidates of Head, Neck, and Thyroid Cancer Cancer Res., March 1, 2005; 65(5): 1693 - 1699. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Weber, M. A. Aldred, C. D. Morrison, C. Plass, A. Frilling, C. E. Broelsch, K. A. Waite, and C. Eng Silencing of the Maternally Imprinted Tumor Suppressor ARHI Contributes to Follicular Thyroid Carcinogenesis J. Clin. Endocrinol. Metab., February 1, 2005; 90(2): 1149 - 1155. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
) and loci with LOH (
) are shown. The area of light gray shading indicates the minimal common region of LOH in the FTC, whereas the minimal common region of LOH in the HCC is shaded in dark gray. Tumor grades and patients’ disease statuses are listed at the top. Tumor grades: 1, 2, 3, and N/A (not applicable). Status: AFD, alive and free of disease; RD, residual/recurrent disease; DOD, dead of disease; DOC, dead of other causes; N/A, not applicable.