Array-based comparative genomic hybridization of colorectal, papillary thyroid, and anaplastic thyroid carcinomas. Comparative genomic hybridization did not reveal gains or losses in indolent papillary thyroid carcinomas, while aggressive colorectal cancers and anaplastic cancer exhibited significant copy number aberrations. Chromosome position is shown on the x-axis; the vertical line indicates the position of the centromere. Representative results for chromosomes 7 (chr7) and 8 (chr8) are shown.
Spectral karyotypes of papillary and colorectal carcinomas. A, Papillary carcinoma: a representative karyotype of a papillary thyroid carcinoma cell. No changes in chromosome number or structure were detected in the metaphases examined from 7 tumors. B, Colorectal carcinoma: translocations detected in several metaphases from a single colon carcinoma. Each block shows a translocation(s) from an individual cell and demonstrates the extensive large-scale genomic instability and substantial intratumoral heterogeneity that typifies these tumors.
Evolution of aggressive tumors is aided by multiple forms of genomic instability. Multiple forms of genomic instability arise sequentially during tumor progression. Early events, detected by inter–simple sequence repeat polymerase chain reaction, are copious but small. Many mutations remain silent and evolution to relatively “benign” neoplasms is enabled. Later events, detected by allelotyping, array-based comparative genomic hybridization, or spectral karyotyping, are larger and unmask previously silent mutations. Selection for aggressive disease occurs.
Stoler DL, Nowak NJ, Matsui S, Wiseman SM, Chen N, Dutt SS, Bartos JD, Loree TR, Rigual NR, Hicks WL, Sait SN, Anderson GR. Comparative Genomic Instabilities of Thyroid and Colon Cancers. Arch Otolaryngol Head Neck Surg. 2007;133(5):457-463. doi:10.1001/archotol.133.5.457
To assess the forms and extent of genomic instability in thyroid cancers and colorectal neoplasms and to determine if such measurements could explain the generally excellent prognosis of thyroid malignant neoplasms compared with colon carcinoma.
Tumor genome analyses. Genomic instability was measured by the following 4 methods, listed in ascending order based on the size of events detected: inter–simple sequence repeat polymerase chain reaction (ISSR-PCR), fractional allelic loss (FAL) analysis, array-based comparative genomic hybridization (aCGH), and spectral karyotyping (SKY).
The genomic instability index of 32 thyroid carcinomas, 59 colon carcinomas, and 11 colon polyps was determined by ISSR-PCR; no difference was seen among the 3 groups by this method. Fractional allelic loss rates were comparable in thyroid cancers and colon polyps and lower than FAL rates in colorectal cancers. Indolent papillary thyroid carcinomas were essentially diploid with no large-scale alterations in chromosome number or structure when evaluated by aCGH or SKY. In anaplastic thyroid cancers, aCGH revealed abundant chromosome alterations. Colorectal carcinomas showed extensive copy number changes and chromosomal rearrangements when analyzed by aCGH and SKY.
Genomic alterations in papillary thyroid carcinoma, such as in benign colon polyps, are principally smaller events detected by ISSR-PCR. With the more aggressive tumor types (ie, anaplastic thyroid and colorectal carcinomas), larger events detected by FAL analysis, aCGH, and SKY were revealed. We hypothesize that mutations caused by smaller genomic alterations enable thyroid cells to achieve a minimal malignant phenotype. Mutations for aggressive biological behavior appear with larger genomic events.
Cancer represents the consequence of somatic evolution, with natural selection of advantageous genomic alterations. This rapid evolution may result from an increased rate of mutation (genomic instability) or from epigenetic events. In either case, tightly controlled cellular processes are dysregulated to bring about the malignant phenotype. Genomic instability has emerged as a key underlying process in the evolution of normal cells to malignant masses, enabling both tumorigenesis and tumor heterogeneity. Previous analyses of colorectal tumors using varied methods have revealed that several independent forms of genomic instability are detectable in these tumors.1- 3 The genomic damage detected by these different techniques range in size from the whole chromosomal, through subchromosomal events, to single base-pair alterations.
The American Cancer Society estimates that the incidence of thyroid cancer in 2006 will be about 30 000 new cases, with an estimated mortality of only 1500 individuals.4 Thyroid cancer encompasses a broad spectrum of histologic types, ranging from well-differentiated papillary and follicular carcinomas to undifferentiated anaplastic tumors. While papillary thyroid tumors, representing the majority of these endocrine cancers, tend to be indolent and associated with excellent long-term patient survival, very rare anaplastic tumors are highly aggressive and patient survival is generally less than a year after diagnosis.5 It has been hypothesized that anaplastic thyroid tumors are the product of continued evolution of well-differentiated carcinomas based on the coexistence of anaplastic and well-differentiated tissues within a single tumor.5,6
In contrast to thyroid cancer, colorectal carcinoma is a far more prevalent and far more aggressive disease. Cancer of the colon or rectum is the fourth most common cancer in the United States, with an estimated 148 000 new cases and 55 000 deaths in 2006.4 The majority of these colorectal cancers will have progressed through a benign polyp stage before achieving malignancy.
In an effort to determine the molecular basis for their very different biological behavior, we compared the patterns of genomic instability previously reported for colorectal carcinoma2,7,8 with those of thyroid cancers. Genomic instability was detected by bacterial artifical chromosome (BAC) array-based comparative genomic hybridization (aCGH), spectral karyotyping (SKY), loss of heterozygosity (LOH), and inter–simple sequence repeat polymerase chain reaction (ISSR-PCR) analyses. Genomic instability as measured by ISSR-PCR has been previously reported.9,10
The 2 study populations, 59 patients with colorectal cancer and 33 patients with thyroid cancer, are described in detail elsewhere.7,10 Specimens were procured under the supervision of the institutional review board of the Roswell Park Cancer Institute, Buffalo, NY, and informed consent for participation in this study was obtained from all patients preoperatively. Histologically normal and tumor tissue from the same patient were procured immediately after surgery. Hematoxylin-eosin–stained sections of paraffin-embedded, neutral-buffered, formalin-fixed (10% formalin by volume in water; pH 7.4) specimens were reviewed by a pathologist to confirm the histologic features of the procured tissue. The tumor and normal tissue were either cultured, as described in the “Spectral Karyotyping” subsection, or stored at −70oC for subsequent DNA extraction for all other genomic assays. DNA extraction from the frozen tumor and normal tissue specimens was carried out as previously described.7
The ISSR-PCR analyses were performed as previously described.7,9 Genomic instability index is the percentage of altered PCR products amplified from an individual's tumor DNA compared with the PCR products of that individual's normal DNA. This method detects alterations that have occurred in normal sequences between 2 microsatellites and is totally distinct from microsatellite instability.1,7
The LOH assays were performed on 26 papillary thyroid cancers as previously described.1 Primers for all 22 autosomes are listed in Table 1. Fractional allelic loss (FAL) was calculated by determining the total fraction of all informative markers for a tumor DNA that showed LOH.1
The details of BAC aCGH were as previously described by Cowell and Nowak.11 Briefly, DNA isolated from tumor tissues and a sex-mismatched pool of 15 normal DNA samples were separately labeled with either Cy3 or Cy5 fluorescent tags, mixed, and competitively hybridized to an array of over 6500 immobilized BAC clone targets. The ratio of red (Cy5) to green (Cy3) signal was quantified and plotted graphically to reveal amplifications and deletions within the tumor genome. Array-based CGH was performed on 16 papillary thyroid carcinomas and 4 anaplastic thyroid carcinomas.
Short-term primary cultures of 7 papillary thyroid carcinomas were established according to the method of Matsui et al.12 Metaphase chromosome preparations were prepared using standard hypotonic treatment and air-drying methods. Chromosomes were sequentially digested with ribonuclease (RNase) and pepsin, as recommended by Applied Spectral Imaging, Carlsbad, Calif, and then denatured in 70% formamide. Human SKY probes representing a mixture of individual chromosome DNA samples were hybridized to the metaphase preparations. Following in situ hybridization and developing chromosome-specific fluorescence spectra with rhodamine, Texas red, Cy5, Cy5.5, and fluorescence isothiocyanate, SKY images were captured using interferometer-based spectral imaging, and karyotypes were prepared using the SKYView software (version 1.62). At least 10 metaphases were examined for each tumor specimen.
Papillary thyroid cancer exhibits moderate clinical behavior compared with other malignant neoplasms such as colorectal carcinoma or anaplastic thyroid cancer; 10-year survival occurs in greater than 90% of patients with a diagnosis of having cancer.13 We investigated if the basis of this difference in clinical behavior might be attributed to the types or degree of genomic instability detected in these tumors. In several previous studies, we have used the genome sampling technique of ISSR-PCR to quantify one form of genomic instability in benign and malignant thyroid lesions and colorectal cancers and polyps.7,8 In addition, we have examined other forms of genomic instability in colorectal neoplasms using allelotyping, BAC aCGH, and/or SKY.1- 3 We have applied the latter 3 techniques to thyroid cancers and compared the types of genomic instabilities detected in indolent papillary thyroid cancer with the more aggressive anaplastic thyroid cancer and colon cancer and with benign polyps.
Genomic instability as measured by ISSR-PCR has been found to be an early event in colorectal carcinogenesis.8,14 Sequence analysis of altered ISSR-PCR products generated from tumor DNA has revealed that this technique preferentially detects small alterations in the genome.15 The mean genomic instability index estimate of genomic damage for 59 colorectal carcinomas was nearly identical to that measured in 7 adenomatous polyps (3.9% vs 4.1%).8 Our 2 subsequent analyses of papillary, follicular, and anaplastic thyroid cancers using ISSR-PCR revealed mean genomic instability index values of 2.9% and 3.1%.9,10 These levels of genomic damage were not statistically different from those seen in colorectal carcinomas or polyps. Furthermore, the genomic instability index of the 1 anaplastic carcinoma available for study was 3.0%, suggesting that a large increase in this form of genomic instability is not responsible for the evolution to the more aggressive tumor type. These data are summarized in Table 2.
Our published study of allelic loss in colorectal cancers and polyps demonstrated a greater than 3-fold higher fraction of markers in carcinomas compared with adenomas; FAL rates were 0.095 in malignant neoplasms vs 0.03 in the benign lesions.1 Allelotyping generally detects allelic loss events of moderate size ranging from less than a megabase to entire chromosomes. The lower frequency of LOH in polyps indicates that these are events more closely associated with the later stages of tumor progression. Using a panel of 21 markers that map to 18 of the autosomal chromosomes, we interrogated the LOH incidence in 26 papillary thyroid cancers (Table 1). Fewer than half of the markers tested demonstrated 1 or more allelic losses. Loss of heterozygosity was detectable in 52% of the tumors, ranging from 0 to 0.154, with a mean FAL of 0.035 (range for colorectal cancers, 0 to 0.450). The mean level of FAL in papillary thyroid cancer was comparable to that observed in benign colon adenomas, far below levels seen in colorectal malignancies.
Bacterial artifical chromosome aCGH is a high-resolution screening technique that assesses DNA copy number aberrations in tumors. An analysis of 33 colorectal carcinomas revealed numerous gains and losses on nearly all chromosomes.3 As might be expected, polyps had significantly fewer chromosomal copy number aberrations, indicating that this form of instability generally occurs later in colon tumor progression.14 Interestingly, papillary carcinomas analyzed by aCGH were again found to be more similar to colon polyps than to carcinomas. The aCGH profiles for these cancers showed no significant deviation from normal diploid DNA (Figure 1). Occasional spikes were observed in one tumor or another, but no consistent alterations in copy number were observed. However, with the evolution to the more aggressive anaplastic thyroid tumor, aCGH detected amplifications and deletions on several chromosomes. Most alterations were only found in a single tumor's DNA and may represent random DNA damage. The deletions and amplifications on the p and q arms, respectively, of chromosome 8 in Figure 1 were detected in 2 of the 4 anaplastic tumors.
To assess levels of genomic instability at the chromosomal level, 7 papillary thyroid cancer primary cultures were subjected to SKY. At least 10 metaphases were scored per tumor. All cells observed in all 7 tumors were scored as diploid with no detectable structural changes. A representative thyroid tumor cell is depicted in Figure 2A. Colon tumors cells were aneuploid, frequently pseudotetraploid, and contained numerous translocations (Figure 2B). Each panel in Figure 2B contains 1 or more altered chromosomes obtained from 7 different cells from the same colon tumor, indicating that colorectal tumors are highly heterogeneous. Anaplastic thyroid tumors were not available for primary cell culture.
The data summarizing all 4 measurements of genomic instability in papillary thyroid, anaplastic thyroid, and colorectal carcinomas, as well as colon adenomas, are given in Table 2. While all of these tumors possess ISSR-PCR–detected instability, those tumors that are associated with better prognoses (ie, papillary thyroid cancer and polyps) lack high-level genomic instability as measured by the remaining methods in Table 2.
Genomic instability, generally accepted as a facilitator of solid tumor progression, occurs in 3 general forms, microsatellite instability, aneuploidy, and intrachromosomal instability. While much has been learned about the underlying mechanisms of microsatellite instability and aneuploidy, less is known of the molecular basis of intrachromosomal instability. Indeed, it now appears that intrachromosomal instability can be subdivided into several independent forms, each presumably with its own molecular origin, that are apparent when different assay methods are used.
In this study, we have compared the genomic instabilities detected in thyroid carcinomas with those observed in colorectal cancers and adenomas to determine if differences in their instability profiles were related to the distinct clinical behavior of the 2 cancers. Four methods were used to measure diverse genomic instabilities that result in genome alterations ranging in size from a few base pairs to whole chromosomes.
Our results demonstrate that thyroid cancers possess levels of genomic instability as measured by ISSR-PCR that are equivalent to those detected in both malignant and benign lesions of the colon or rectum. In addition, ISSR instability in anaplastic carcinoma was no greater than in papillary cancer. Therefore, this form of instability is not likely to be responsible for the evolution of aggressive behavior observed in the anaplastic carcinoma.
A meta-analysis of published studies of LOH in thyroid cancer performed by Ward et al17(p525) revealed that “papillary carcinomas had exceedingly low rates of LOH.” In 2 separate genome-wide allelotyping studies, Vogelstein et al18 and Califano et al19 confirm that FAL levels (defined in both studies as the fraction of chromosomal arms lost) in papillary thyroid cancers are far lower than those of colorectal cancer. In addition, anaplastic thyroid cancer has been shown to exhibit numerous LOH events on several different chromosomes.16
We were unable to find any reports of aCGH analysis of thyroid malignant neoplasms for comparison with our data. Our aCGH analysis detected few gains or losses in papillary thyroid carcinoma relative to the anaplastic form of the disease or to colon cancer. In addition, we did not detect any consistent changes. Several investigators have used metaphase-based CGH to analyze thyroid cancers. Bauer et al20 reported that only 4 of 15 papillary thyroid cancers exhibited less than 4 aberrations per tumor, with the remaining 11 having normal diploid karyotypes. Wreesmann et al21 found few gains or losses in 15 well-differentiated papillary thyroid cancers and numerous alterations in 27 poorly differentiated or anaplastic carcinomas. In a subsequent report from this group, higher numbers of gains and losses were reported in follicular variant of papillary carcinoma, but very few alterations were detected in classic papillary cancers.22 Miura et al23 demonstrated several copy number aberrations in anaplastic tumors and an anaplastic cell line, but only 2 copy number aberrations in the papillary cell line they evaluated. Gains of chromosome 8q were reported as a recurrent event in anaplastic thyroid cancers and cell lines by Wilkens et al.24 The low level of aCGH-detected genomic instability in papillary thyroid cancer indicates that the events measured by ISSR-PCR in this disease are smaller than the 150 000–base pair average size of BAC clones. Furthermore, our data coupled with the many published reports, suggest that large-scale alterations of the genome are typical of progression to anaplastic carcinoma.
To our knowledge, this study is only the second report of SKY of nonradiation-induced human thyroid malignant neoplasms. Our analyses did not detect any numerical or structural aberrations at the chromosomal level. In the report by Foukakis et al,25 only 1 translocation in 1 of 10 papillary thyroid cancers assessed was detected. Together, our data suggest that large-scale chromosomal events are rare in papillary thyroid cancers. Also, these tumors did not exhibit the type of heterogeneity we observed using SKY in colon cancers. Foukakis et al25 also reported karyotype data from a single anaplastic thyroid cancer cell line. Abundant chromosomal aberrations were observed using SKY coupled with standard G-banding, including a high-level amplification of chromosome 8q23-qter. Whether these chromosomal abnormalities and tumor heterogeneity exist in anaplastic carcinomas remains to be determined.
Our measurements of genomic instability and those from the literature have led us to the model of genomic instabilities in tumor progression presented in Figure 3. Several forms of genomic instability can occur within a tumor cell, some arising earlier than others.
The earlier wave of genomic damage consists of thousands of small genome alterations that are readily detected by ISSR-PCR,8,10 not as readily detected by allelotyping, and rarely detected by assays such as aCGH or SKY (Figure 3A).
Because these events are small, the probability of hitting both copies of a particular sequence is low and the effects of some mutations remain masked. Those tumors in which this type of instability predominates tend to be relatively benign (ie, adenomas and papillary thyroid carcinomas). Subsequent waves of genomic instability tend to cause larger damage to the genome in forms such as deletions, amplifications, and uniparental conversion of some chromosomes, which can be revealed by allelotyping, SKY, and aCGH.
With the onset of large event–causing instabilities, the probability of revealing a previously silent mutation increases. Unmasked mutations are now able to contribute to the evolution of the tumor to more aggressive disease.
The tumors that exhibit multiple forms of genomic instability are clinically more aggressive tumors such as colorectal cancer and anaplastic thyroid cancer. The combination of these multiple forms of instability enable normal cells to evolve through benign neoplasms to highly aggressive malignant neoplasms.
Correspondence: Garth R. Anderson, PhD, Department of Cancer Biology, Roswell Park Cancer Institute, Buffalo, NY 14263 (firstname.lastname@example.org).
Submitted for Publication: July 31, 2006; accepted November 7, 2006.
Author Contributions: Drs Stoler and Anderson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Stoler, Nowak, Loree, and Anderson. Acquisition of data: Stoler, Nowak, Matsui, Wiseman, Chen, Dutt, Bartos, Loree, Rigual, Hicks, and Sait. Analysis and interpretation of data: Stoler, Nowak, Wiseman, Bartos, Loree, Rigual, and Anderson. Drafting of the manuscript: Stoler and Hicks. Critical revision of the manuscript for important intellectual content: Nowak, Matsui, Wiseman, Chen, Dutt, Bartos, Loree, Rigual, Sait, and Anderson. Statistical analysis: Bartos. Obtained funding: Loree and Anderson. Administrative, technical, and material support: Stoler, Nowak, Matsui, Wiseman, Chen, Dutt, Bartos, Loree, Rigual, Hicks, Sait, and Anderson. Study supervision: Anderson.
Financial Disclosure: None reported.
Funding/Support: This work was supported by grants RO1 CA 74127 (Dr Anderson) and P30 CA16056 from the National Institutes of Health.
Previous Presentation: This work was presented at the American Head and Neck Society 2006 Annual Meeting and Research Workshop on the Biology, Prevention, and Treatment of Head and Neck Cancer; August 17-20, 2006; Chicago, Ill.
Acknowledgment: We acknowledge Ann Marie Bauer, RN, MSN, for invaluable assistance in the Head and Neck Surgery clinic at Roswell Park Cancer Institute, Buffalo, NY.