—  SHORT COURSE #12  —

Molecular Analyses in Endocrine Pathology
Dr. George Kontogeorgos
Dr. Robert Yoshiyuki Osamura
Dr. Jennifer Hunt

Section 6 - Technique: Loss of Heterozygosity

Jennifer Hunt
The Cleveland Clinic
Cleveland, OH


Tumor Suppressor Genes

Tumorigenesis through the inactivation of tumor suppressor genes was first postulated by Dr. Alfred George Knudson in the early 1970's, in seminal work with retinoblastoma [1, 2]. Knudson elegantly showed that hereditary retinoblastoma and sporadic tumors had similar molecular mechanisms that involved the retinoblastoma gene. His theory, which is now known as Knudson's Hypothesis, predicted that both copies of the retinoblastoma gene needed to harbor mutations for tumorigenesis to occur.

Normal cells have two functional copies of each wild-type (non-mutated) tumor suppressor gene, one inherited from each parent. In the process of tumorigenesis, one copy of the TSG develops an inactivating mutation. This first mutation is often a point mutation or small deletion mutation, and this is termed the first genetic hit. This first hit can occur spontaneously or it can be inherited in cancer syndromes that make patients susceptible to specific tumors. Alone, the first mutation is not tumorigenic and usually has no ramifications for cell function, because the second copy of the TSG is still functioning. This is why tumor suppressor genes are also referred to as "recessive genes"—they require both copies to be mutated in order for loss of function [3]. The second genetic hit affects the second copy of the TSG, and this may be due to point mutations or more commonly, larger deletion mutations. It is uncommon for cells to have two large deletion mutations (biallelic deletion), because these would be considered to be lethal for the cell [4]. The cells that harbor these two mutated copies of the TSG have a loss of function of the TSG activity, and this is how carcinogenesis is thought to be stimulated.

One of the most famous illustrations of Knudson's hypothesis is the Vogelstein model of familial colon cancer progression [5, 6]. In this classic example, germline mutations in the Familial Adenomatous Polyposis gene (APC, chromosome 5q21) are the first hit, and somatic mutations (usually deletions) in the second copy of the APC gene represent the second hit. With both copies of the gene functionally lost, colon cancer can proceed. The colon adenoma to carcinoma pathway is probably one of the best studied examples of tumorigenesis via the tumor suppressor pathway. Some of the most common tumor suppressor genes are listed in TABLE 1.

TABLE 1: Common tumor suppressor genes, their chromosomal location, function, and any associated inherited syndromes

Tumor suppressor gene Chromosomal location Function Inherited syndromes
P53 17p13 Cell cycle regulation, apoptosis Li-Fraumeni Syndrome
RB1 13q13 Cell cycle regulation Retinoblastoma
WT1 11p13 Transcriptional regulation Wilm's tumor
NF1 17q11.2 Catalysis of RAS inactivation Neurofibromatosis Type 1
NF2 22q12.2 Linkage of cell membrane to cytoskeleton Neurofibromatosis Type 2
APC 5q21 Signaling through adhesion molecules Familial Adenomatosis Polyposis
DCC 18q21.3 Transmembrane receptor
BRCA1 17q21 Repair double strand breaks Familial breast cancer
BRCA2 13q12.3 Repair double strand breaks Familial breast cancer
MSH2 2p22-p21 DNA mismatch repair Mismatch repair syndromes
MLH1 3p21.3 DNA mismatch repair Mismatch repair syndromes
VHL 3p26-p25 Regulation of transcription elongation Von Hippel Lindau syndrome
PTEN 10q23 Regulated cell survival Cowden Syndrome

Tumor suppressor gene mutations and loss are relatively specific to neoplastic and pre-neoplastic conditions [7], but are not mandatory. Tumorigenesis involves many different alternate pathways, including oncogene mutation, defects in DNA mismatch repair, DNA methylation, and others. Some investigators have suggested that models predict that there are not enough random mutational events to account for carcinoma in the high rates that we see it in the general population. They suggest that there must be a "mutator" genotype, which is to date completely unidentified [3]. This mutator genotype would increase mutation rates in a non-random manner, and make individuals susceptible to cancer. Other investigators have proposed that all neoplasia is initiated by aneuploidy, which sets the cell up for propagation and expansion of non-lethal and lethal mutations [8].

We face many challenges in assessing the genetics of tumors. First, the genotypes and karyotypes seen in carcinoma are complex, and are the result many different intersecting pathways. The competing theories on tumorigenesis remain to be resolved. . Second, we face issues in our ability to set up controlled experiments using human tissues since resources and quality of materials may be limited. Finally, the assays that are utilized in the literature are diverse and do not use common platforms or interpretation guidelines. The following syllabus will cover the theory behind and the details of performing two common tumor assays: loss of heterozygosity analysis and microsatellite instability analysis.

Chromosomal or Genomic Instability

Although we think of DNA as being very stable, replication errors are extremely common, even in entirely normal cells [9]. And, we all have exposures to carcinogens on a regular basis (UV exposure being the most universal), and these may induce even more genetic damage. In normal cells, DNA damage and replication errors may be either corrected through the DNA repair system, or if the damage is severe enough, the cells will undergo spontaneous apoptosis.

There are many theories about how carcinogenesis is initiated. Two commonly cited theories are tumor initiation via oncogene mutations and tumor initiation via aneuploidy [10]. In both theories, chromosomal instability is an important part of tumor development. It is thought that the instability is an early part of tumorigenesis and that some of these early mutational events are the initiating forces for tumor development [11]. Chromosomal instability can result in different karyotypic or genomic changes, most of which can be measured with specific assays.

Cells themselves can be individually examined for chromosomal alterations using either cytogenetics or FISH type assays. These assays play an important role in molecular pathology. The DNA from an area within a tumor can also be examined using polymerase chain reaction (PCR) based assays, which enable us to see clonal changes in the population. For most PCR-based assays, there is a lower limit of detection for the sensitivity and it varies based on the assay. Some quantitative assays have a very low limit of detection (even down to 1% aberrant DNA in a background of normal DNA). Most assays used in anatomic pathology, however, will need enriched tumor cell populations.

The enrichment of DNA is necessary to exclude normal cell DNA from the sample. This is most commonly accomplished by microdissection of samples, whether they are frozen sections or paraffin embedded samples. Microdissection can be done by hand using simple laboratory equipment, or can be done using machinery for laser capture microdissection [12]. It is important to remember that even with microdissection, the DNA from thousands of cells would be assessed all together. Minor changes or changes in small populations will likely not be picked up by most PCR based assays. For example, a point mutation can only be seen by traditional sequencing reactions if >20% of the cells in the population harbor the mutation. Therefore, our PCR-based assays are best at detecting clonal DNA damage, or DNA mutations that are propagated.

We do know that the DNA of tumors will replicate, despite having mutational damage, and is more likely to propagate the DNA mutations to progeny cells without monitoring and correcting the defects. This may be one of the reasons that tumors tend to accumulate more and more mutations in a stochastic manner over time as a tumor develops and grows. The most common type of DNA damage will include point mutations, deletions, and amplifications of specific genes or genetic areas [3, 13]. It is likely that these errors and damage occur throughout the genome, but are mostly recognized and measured at hotspots that are either associated with tumorigenesis or are particularly vulnerable to errors in the first place. Hotspots for deletion mutations would commonly include tumor suppressor genes. The measurement of tumor suppressor gene loss is usually accomplished though an assay measuring "loss of heterozygosity (LOH)".

Gene Loss and Loss of Heterozygosity

DNA Polymorphisms
A huge percentage of the genetic code is redundant, with up to 99% homology between humans and chimpanzees, and up to 99.9% homology between individuals [14, 15]. However, each human being still has a unique DNA genetic code, which accounts for major phenotypic differences. The differences in the genetic code are predominantly due to variable regions, which can be single nucleotide differences or repeat length differences. These variations between individuals are called polymorphisms. Certainly, the value of identity testing and genetic identification is clear, especially in highly publicized court cases in which DNA is used to link a crime with a suspect or to clear a suspect of a crime that he or she did not commit [16, 17, 18, 19]. Another application of this type of testing is in paternity testing, in which genetic polymorphisms are traced back one generation [20, 21].

There are several different types of variable areas with in the genetic code. The most common and most utilized for genetic testing are the short tandem repeats (STR) and the single nucleotide polymorphisms (snp). STRs are short redundant nucleotide sequences that vary from 2 to 7 base pairs in length that are repeated for a variable number of times. An example of a tetranucleotide repeat sequence (4 basepair repeat) is shown in FIGURE 1.



When the polymorphism is just one base and there is no disease associated, the polymorphism is called a single nucleotide polymorphisms (snp, pronounced "snip"). Single nucleotide polymorphisms generally have two variants in the population, one of which is usually more common. As the name predicts, a snp is a location in the genome at which there is variability. On one public/private website, there are currently over 1.8 million catalogued SNPs (http://snp.cshl.org/ ). An example of a snp is given in FIGURE 2 below.



When an individual has inherited two copies of the polymorphism that are different from one another (have different numbers of repeats or a different base at the point of a snp), they are said to have an informative genotype at that locus. This informativeness translates into heterozygosity for the polymorphism (two different copies) and this means that their two copies of the polymorphism can be discriminated from each other by PCR-based assays. If an individual has two copies of the gene with the same number of repeats on each copy or the same snp basepair, then that locus is non-informative. The PCR products from these patients at the site of the polymorphism will be identical.

There are important applications of testing for STRs and SNPs for identify testing both in forensic and clinical medicine. However, because the rates of polymorphism for STRs vary, depending upon the population being studied, panels of polymorphisms are always used for identity testing. For criminal identity testing, especially when genotypes are matched, a statistical calculation is performed to indicate the probability that two individuals would exist with identical genotypes [22]. In anatomic pathology, we are more concerned with the opposite situation, when two unmatched samples are discovered, such as in paraffin embedded tissue blocks or specimen mix-ups [23, 24].

Loss of Heterozygosity Analysis
Loss of heterozygosity (LOH) analysis relies upon our ability to discriminate between the two inherited copies of a particular gene. Unfortunately, the coding regions of our genes are usually highly conserved through evolution and are therefore identical in sequence on each of the two alleles that we have inherited (one maternal allele and one paternal allele). So the challenge in looking for loss of one copy of a gene is actually in being able to tell the two copies apart from one another. Technologies have evolved to do this type of assessment that utilize the polymorphisms in our genome. These polymorphisms occur frequently in the genome, with an estimate periodicity of at least 1 polymorphism per 1000 basepairs. We can almost always find polymorphisms that are in close proximity to genes of interest. All of these polymorphisms do suffer from a common problem: they are not 100% variable, and there will still be some individuals that, by chance, inherited the same sequence in the variable region from both the mother and the father.

PCR is used in an LOH analysis, with primers that flank the STR. In an informative person at that locus, the PCR amplicons will have different lengths since they have two copies of the gene that have different numbers of repeat units. In the example in FIGURE 3, the PCR product from the first copy will be have a 181 basepair product (with 11 repeat units of a tetranucleotide repeat) and the second copy will have a 197 basepair product ( with 16 repeat units of a tetranucleotide repeat).



PCR products of different sizes will migrate at different speeds, when exposed to electric current in electrophoresis. In normal cells, if we analyze an STR locus in an informative patient, we expect that there will be two differently sized PCR products of approximately the same amount, one from each chromosomal copy of the gene. In tumor cells, there may be deletion mutation present, which will alter the ratio of the PCR product amount from the two different PCR products. When one copy of the STR is lost, there will be only one PCR product, or the genotype will appear to be homozygous at that locus. Because the normal was originally heterozygous, we label this situation as "loss of heterozygosity". A ratio of the amount of PCR product present for the two alleles can be obtained from capillary electrophoresis electropherograms as the ratio of the peak heights for the two products. If one is using gel based electrophoresis, this assessment is by visual inspection—if one band is less than 50% the intensity of the second band, this is indicative of loss. The calculation to determine if there is true loss of genetic material should include a comparison of the tumor allele ratio to that of the normal allele ratio. This will help to account for variability in PCR for the two alleles that might be secondary to the size of the PCR products or other variables that affect amplification efficiency [25].

Pitfalls and Problems with LOH Assessment

Poor Markers
It is critical to realize that measuring the amount of a polymorphic repeat unit that is present near a tumor suppressor gene is only a surrogate marker for that gene itself. There are several different variables that will affect how tightly the polymorphism is linked to the tumor suppressor gene. The most important of these will be the distance between the TSG and the polymorphism. If a polymorphism is very far from the TSG of interest, it will not represent an accurate measurement of the TSG. These data are rarely provided in the literature, but can be searched in on-line databases.

Low DNA Concentrations
Another major problem in LOH analysis can arise if there is inadequate DNA [26]. When the DNA concentration is very low, one allele may be dramatically preferentially amplified over the other. Then, the final assessment of the allele ratio will not be reflective of the original amount of starting DNA of each allele. Whether the starting DNA concentration is affecting the allele ratio can be assessed by repeating samples or running samples in duplicate. By seeing more than one reaction with a similar or very close ratio, one can be reassured that the allele ratio truly reflects the amount of starting template from each allele.

Lack of a Normal Comparison Sample
Sometimes in tumor pathology, we do not have the luxury of a concordant normal sample from the patient. This is particularly true when analyzing gliomas, since normal brain may not be included in the biopsy sample. It is always preferable utilize normal that has undergone identical tissue processing procedures. But, when this is not available, other normal tissue can be requested. This would include the most common specimens, such as blood, but can also include a simpler specimen to obtain, the buccal brush. Even unique specimens like hair and fingernail clippings can be used for this type of assessment. The allelic dropout rates tend to be higher in these latter specimens, however, since the DNA is in low concentration.

CASE ONE - PARATHYROID

Case History:
A 50 year old female presented with symptomatic hyperparathyroidism. A large parathyroid was discovered on sestamibi scans. The surgery was uneventful, with the exception of the gland being adherent to the thyroid gland. An en bloc resection was performed with a parathyroidectomy and a thyroid lobectomy.

Pathology:
The gland weighed 1054 mg. On cut surface, it was gray, with some indication of fibrosis. A frozen section was performed and showed significant fibrosis. The frozen section diagnosis was deferred. Histologically, it was hypercellular and showed predominantly chief cells. There were broad bands of fibrosis, but no increase in mitotic activity or nuclear pleomorphism. One focus of vascular invasion was seen.

Molecular : The molecular mutational results are shown:

MARKER GENE LOCUS TUMOR
1s1161 1p 1p35.1 LOH
1s407 1p 1p36.21 LOH
1s461 HPT-JT 1q21 LOH
1s384 HPT-JT 1q21 NI
3s1516 VHL 3p25.3 LOH
3s1539 VHL 3p25.3 NI
3s1600 VHL 3p25.3 NO
5s659 APC 5q23.2 NI
5s1384 APC 5q23.2 LOH
7s486 MET 7q31.3 NO
9s251 p16 9p21.3 LOH
10s520 PTEN 10q23.31 LOH
10s1171 PTEN 10q23.31 LOH
10s1173 PTEN 10q23.31 LOH
D12S375 MDM2 12q21.1 NO
13s319 RB 13q14.3 LOH
13s1319 RB 13q14.3 NI
13s4177 RB 13q14.3 LOH
17s516 p53 17p13.1 LOH
17s1844 p53 17p13.1 LOH
17s1877 NM-23 17q21-23 LOH
22s1150 NF2 22q12.2 LOH
OVERALL LOH RESULTS 15/18
FAL CALCULATION 83%

Discussion of Case 1
Hyperparathyroidism is relatively common with some estimates of incidence in European countries of 3-4 per 1,000 [27]. Comparable numbers would be expected in the United States [27, 28] . Causes of hyperparathyroidism are primary, secondary, or tertiary (TABLE 2) [27, 29] . Patients with hyperparathyroidism can be asymptomatic with hypercalcemia identified on routine health screens, or they may be very symptomatic with an increased risk of hypertension, nephrolithiasis, osteoporotic fractures or cardiovascular complications [29, 30, 31] .

TABLE 2: Causes of hyperparathyroidism

Type Description Causes
Primary Hyperparathyroidism Parathyroid related disease
  • Parathyroid adenoma

  • Parathyroid hyperplasia (genetic)

  • Parathyroid carcinoma
Secondary Hyperparathyroidism Secretion of parathyroid hormone in response to low calcium from another disease
  • Rickets (Osteomalacia) due to vitamin D or calcium deficiency)

  • Sprue

  • Chronic renal failure
Tertiary Hyperparathyroidism One parathyroid gland becoming autonomous after persistent secondary hyperparathyroidism



Some forms of hyperparathyroidism are associated with genetic syndromes. These include the familiar disease that is associated with the multiple endocrine neoplasia (MEN) syndromes, and more rare forms, such as that seen in the uncommon disease Hyperparathyroidism-jaw-tumor syndrome (HPT-JT). The different types of familial hyperparathyroidism are listed in TABLE 3.

TABLE 3: Different types of familial hyperparathyroidism

Syndrome Name Pattern Gene Locus Clinical Features
MEN Type 1 (MEN1) Dom MEN1 11q13 Multiglandular PT disease (>90%)
Gastroenteropancreatic tumors
Pituitary adenomas
MEN Type 2A (MEN2A) Dom RET 10q21 Multiglandular PT disease (20-30%)
C cell tumors of thyroid
Pheochromocytomas
Hyperparathyroidism jaw tumor syndrome (HPT-JT) Dom HRPT2 1q21-32 Primary hyperparathyroidism with cystic parathyroids
Parathyroid carcinoma (10-15%)
Fibro-osseous lesions of jaws
Kidney lesions
Familial isolated hyperparathyroidism (FIHP) Dom MEN1
CaSR 		11q13
19p13.3 		Benign multiglandular PT disease
Carcinomas of breast, colon, endometrium and other
Familial hypocalciuric hypercalcemic (FHH) Dom CaSR 3q21.1
19p13.3 (type 2)
19q13 (type 3) 		Normal or increased calcium
Moderate hyperphosphatemia
Inappropriately low urine calcium
Increased or normal PTH levels
Neonatal severe hyperparathyroidism (NSHPT) Rec CaSR 3q21.1 Homozygous form of FHH
Autosomal dominant mild hyperparathyroidism
(ADMH) 		Dom CaSR 3q21.1

CaSR = calcium sensing receptor gene, Dom=autosomal dominant, Rec=recessive

The treatment of most types of hyperparathyroidism is usually surgical [32, 33]. There are several known risks of surgery, including damage to the recurrent laryngeal nerve and permanent hypoparathyroidism [34]. Historically, abnormal or enlarged glands are surgically excised, the weight is obtained, and frozen sections are performed to confirm the presence of parathyroid tissue [35, 36]. In cases of adenoma or carcinoma, shave biopsy of at least one normal sized gland has been routinely practiced to rule out the possibility of unrecognized heterogeneous hyperplasia, which can result in operative failure and/or recurrent hyperparathyroidism.

Treatment paradigms have shifted somewhat in recent years, with the more widespread use of rapid, intraoperative parathyroid hormone measurement [37, 38]. In today's practice, patients with single gland abnormalities may undergo resection of only the enlarged gland if intraoperative parathormone measurements are used to functionally exclude hyperplasia [39]. Intraoperative parathyroid hormone levels are obtained before and after removal of an abnormal gland (13, 31) and if the hormone level drops within certain strict criteria, the surgery can be safely concluded. False positive drops in intraoperative parathyroid hormone levels are relatively rare [40]. Frozen sections may still be used to confirm tissue diagnosis. Oil-red-O staining can be performed on non-fixed frozen tissue sections to demonstrate the decrease in intracytoplasmic lipid as compared to staining on the biopsied normal gland, but this finding is characteristic of both hyperplasia and adenoma and this is not usually performed intraoperatively [41, 42, 43].

Adenoma and Carcinoma
The majority of patients with primary hyperparathyroidism will have a single glandular abnormality. These patients usually present with incidental hypercalcemia, found at routine health screenings. A sestamibi scan can help the surgeon to pre-operatively locate the abnormal gland, but these may be nonspecific and thus exploration during surgery may still be needed. In the setting of hyperparathyroidism, a single enlarged parathyroid gland will be surgically excised, in a conservative operation in order to not harm the recurrent laryngeal nerve or other critical structures in the neck. The other glands may be explored, particularly if there is a concern of the possibility of either double adenoma or asynchronous hyperplasia.

For many years, there was an argument that double adenoma did not exist and that if two glands were affected, it was by definition hyperplasia (probably asynchronous type). Today, however, several investigators have provided solid evidence from long-term follow-up of patients with two enlarged hypercellular glands and no recurrence of hyperparathyroidism [44, 45]. This evidence supports the fact that double adenomas can exist, though this is still an uncommon occurrence. .

The first step in assessing a parathyroid gland it the determination if the gland is enlarged. Most people use the weight as the most accurate measurement of the size of a parathyroid gland [46]. The normal weight of the parathyroid gland should be less than 60 mg. But, in an autopsy study, normal glands had a median weight of 26 mg (range 8 to 75 mg); lower weights were seen in patients with chronic diseases [47]. Some investigators have also used the size of the gland, though standard measurements have not been nearly so well established [48].

Parathyroid adenomas are almost always enlarged by weights (i.e. > 60 mg). The mean weight of a parathyroid adenoma is around 500 mg (range 55 to >3000 mg) [46]. The parathyroid adenoma will have typical histologic findings. The cellular composition may include show predominantly a single cell type, or can show mixed features [46]. The most common type of cell to predominate is the chief cell, though oncocytes can also be dominant [49, 50, 51]. Occasionally parathyroid adenomas designated as "water clear cell adenomas" have been described in the literature. The cells in these lesions are composed of polygonal cells with very clear cytoplasm and distinct cytoplasmic borders [52]. Parathyroid hormone levels may be very low in these lesions [53]. Other morphologic features of parathyroid adenoma may include a rim of normal somewhat suppressed parathyroid tissue around the outside of the gland. This can be used as a diagnostic clue for the etiology of the pathologic process. Mitoses are usually rare to absent in parathyroid adenomas [54, 55].

The differential diagnosis between benign adenoma and carcinoma cannot be reliably made just based on the histologic assessment alone (TABLE 4). The surgeon's intraoperative opinion about adherence to surrounding structures is an important criterion in distinguishing benign from malignant parathyroid neoplasms. The surgeon who encounters a parathyroid carcinoma will describe the gland as "sticky," "fibrotic," "hypervascular," or "adherent to local structures" [56, 57]. These descriptions should immediately alert the pathologist to the possibility of parathyroid carcinoma. The histologic features of malignancy may not all be seen in a given case [55, 58]. There are some histopathologic features which are associated with malignancy, though they are certainly not pathognomonic for carcinoma. Worrisome features include the presence of increased or atypical mitoses, broad bands of fibrosis, trabecular growth pattern, invasion of adjacent tissue, and perineural or angiolymphatic invasion [59]. These features usually correlate with malignancy, though these histologic features are not always present in every case of parathyroid carcinoma [55, 60, 61]. Parathyroid carcinomas tend to be locally invasive and invasion is most commonly seen into the thyroid gland, strap muscles, recurrent laryngeal nerve, esophagus or trachea [55, 62].

TABLE 4: Clinical and histologic features of parathyroid carcinoma
Clinical Features Histologic Features
High calcium level (>14 mg/dl) Trabecular growth
Parathyroid hormone level > 5 x normal Broad intersecting fibrous bands
Palpable mass lesion Increased mitoses
Bone symptoms Stromal invasion
Operative findings of invasive growth (sticky, fibrotic, vascular gland) Angiolymphatic or perineural invasion



In the final analysis, it may be difficult to make a diagnosis of parathyroid carcinoma. In some cases, the clinical impression of parathyroid cancer does not correlate with the pathologic or microscopic impression. Therefore these cases are often labeled as "atypical adenoma", and because of diagnostic ambiguity, the true diagnosis may only be resolved with long term follow-up [55, 63] .

For many years, parathyroid carcinoma was thought of as a lethal disease with a terrible prognosis. This prognosis was probably partially related to non-uniform treatment and incomplete excision in some patients. The treatment of choice for parathyroid carcinoma is an en bloc resection with clear margins [58]. Most series of patients with parathyroid carcinoma with optimal treatment have shown recurrence rates of <10% and 5-year survivals of nearly 90% (~65-70% survival at 10-years) [62, 64] . Part of the reason behind the misconceptions about prognosis is the fact that some cases of parathyroid carcinoma have minimal invasion and very few other features of malignancy. These probably represent the lowest grade form of the disease, but no grading system is in place. At the other end of the spectrum will be rare patients who have aggressive malignancies with widespread invasion. One recent report suggested classifying minimally invasive tumors as low grade and widely invasive tumors as high grade [65].

The differential diagnosis for an atypical parathyroid lesion with histologic features worrisome for malignancy can include parathyroid adenomas or cyst that undergo degenerative changes from rupture or trauma; sometimes these features are also seen in the re-operative setting or in persons with neck surgery for other reasons. The presence of hemosiderin and degenerative changes can be helpful in making this diagnosis [55]. Parathyroid cyst rupture can be accompanied by an unusual clinical finding of hematoma and subsequent severe skin bruising [66]. Finally, in patients who have primary parathyroid surgery with spillage during the operation, parathyromatosis can develop [67]. In this condition, remnant parathyroid nests and clusters can be located throughout the neck tissues [68].

Mutational Analysis of Parathyroids
There have been some studies of the molecular mutational findings in parathyroid neoplasia. One feature that is consistently noted is that parathyroid adenomas and carcinomas have a high rate of loss of the short arm of chromosome one (1p) [69, 70, 71, 72, 73]. This is not a feature that is generally seen in parathyroid hyperplasia. Other genes have also been implicated in parathyroid adenomas and carcinomas, including Retinoblastoma (RB, 13q14.3), the MEN gene (11q13), and the BRCA2 gene (13q12.3) [74, 75, 76]. Some studies indicated that protein expression of RB could be used to assess for malignancy. In carcinomas, RB expression is usually lost, and this was originally thought to be secondary to tumor suppressor gene inactivation [77]. However, more recently a detailed analysis of the RB gene did not show any inactivating mutations [78]. There is loss of heterozygosity of RB in carcinomas [79]. Although the definitive involvement of RB in carcinogenesis remains unclear, loss of RB protein expression is seen in the majority of parathyroid carcinomas [80] and at the DNA-level, alterations and loss of heterozygosity of the RB gene are seen in the majority of parathyroid carcinomas [74, 76, 77].

Studies of a very interesting syndrome (hyperparathyroidism—jaw tumor syndrome, HPT-JTS) have also provided insight into the pathogenesis of parathyroid neoplasia. A tumor suppressor gene has been implicated, mapping to 1q25-31 and designated as HPRT2 that harbors germline mutations in hereditary cases of this syndrome. Loss of heterozygosity and somatic point mutations have also been detected in sporadic parathyroid carcinomas [81, 82]. This gene has a lot of promise for future understanding of parathyroid disease.

Another way to assess tumorigenesis is to examine lesions for genomic loss mutations across a broad spectrum of tumor suppressor genes. In many different tumors, this type of analysis demonstrates that the number of genomic micro-deletions (measured as the fractional allelic loss) correlates with increasing degrees of malignancy. Intuitively, it makes sense that tumors that are more malignant will display more mutational damage than tumors that are benign or even non-neoplastic. The panel of tumor suppressor genes that are selected for this type of analysis is usually designed to include the genes that have been studied in isolation and found to have high levels of loss in carcinomas of the organ system. In the parathyroid, we previously demonstrated that this type of analysis is able to separate carcinomas from adenomas and hyperplasias [79]. In fact, histologically and clinically defined malignant tumors had a mean fractional allelic loss of 63%, as compared to benign disease where adenomas had a mean FAL of 11% and hyperplasias had a mean FAL of 15% [83].

Although the molecular mechanisms of carcinogenesis have not been fully worked out, there is the possibility that the type of mutational analysis described above can be implemented at the clinical level. This type of analysis may prove quite useful for lesions designated as "atypical adenomas", which will usually fall out on either side of the spectrum when analyzed at the molecular level. Some investigators have suggested that a molecular approach may become the new standard of care for defining parathyroid neoplasia [58].

CASE TWO - THYROID

Case History:
This is the case of a 55 year old male with a symptomatic large goiter. On ultrasound, it were several solid masses, the largest of which was 4 cm in greatest diameter. An FNA was indeterminate (Hurthle cell lesion). The patient went to surgery and underwent a total thyroidectomy.

Pathology:
Grossly, there were three physically separated encapsulated lesions; one in each lobe and one in the isthmus. The lesions were well sampled histologically. All were nearly identical in morphology and showed Hurthle or oncocytic features. Each lesion had definite vascular invasion at the level of the capsule.

Molecular :
Each tumor was analyzed for a panel of tumor suppressor genes for loss of heterozygosity (LOH). The results are shown:


MARKER GENE LOCUS Tumor 1 Tumor 2 Tumor 3
D1S187 NRAS 1p13.2 LOH T LOH T LOH T
D1S2687 NRAS 1p13.2 LOT B LOH B LOH B
D1S2881 NRAS 1p13.2 LOH B LOH B LOH B
D1S1161 p21 1p35.1 LOH T LOH T LOH T
D3S1516 VHL 3p25.3 LOH B LOH B LOH B
D3S1539 VHL 3p25.3 LOH T LOH T LOH T
D9S251 p16 9p21.3 LOH T LOH T LOH T
D9S1748 P16 9p22.2 LOH B LOH B LOH B
D9S1679 P16 9p22.2 LOH T LOH T LOH T
D9S1851 PTCH 9q22.3 NO LOH NO LOH NO LOH
D10S1171 PTEN 10q23.31 NI NI NI
D10S1173 PTEN 10q23.31 NO LOH NO LOH NO LOH
D10S1178 PTEN 10q23.31 NO LOH NO LOH NO LOH
D10S520 PTEN 10q23.31 LOH B LOH T LOH T
D11S1344 KAI1 11p11.2 LOH T LOH B LOH T
D11S1385 KAI1 11p11.2 LOH B LOH T LOH T
D11S1319 KAI1 11p11.2 NO LOH NO LOH NO LOH
D17S516 p53 17p13.1 NI NI NI
D17S768 P53 17p13.1 NO LOH NO LOH LOH T
D18S1119 DCC/DPC4 18q21.2 NO LOH NO LOH NO LOH
D18S487 DCC/DPC4 18q21.2 NI NI NI
OVERALL LOH RESULTS 12/18 12/18 13/18
FAL CALCULATION 67% 67% 82%



Discussion of Case 2
Follicular derived tumors of the thyroid gland have been traditionally diagnosed by the cellular composition, with follicular carcinomas being reserved for tumors with typical follicular epithelial cells and Hurthle cell carcinoma being the terminology for follicular patterned lesions with oncocytic cellular morphology. This classification system was utilized because of reported differences in behavior and presumed differences at the molecular level. However, in the most recent edition of the WHO classification, Hurthle cell lesions were reclassified as variants of follicular tumors: "Oncocytic variant of follicular carcinoma". For the purposes of the discussion presented here, the older terminology will be utilized, though the tumors will be considered together.

Follicular carcinomas (FCC) and Hürthle cell carcinomas (HCC) are relatively uncommon tumors of the thyroid gland that are derived from follicular epithelial cells and do not have evidence of papillary differentiation. The most common tumors are well differentiated and are typical encapsulated tumors with local invasion; these tumors have a fairly good prognosis [84]. The tumors are categorized based on the level of invasion, with "minimally invasive follicular/Hürthle cell carcinoma" showing capsular invasion alone and "encapsulated angio-invasive follicular/Hürthle cell carcinoma" demonstrating vascular invasion at the level of the tumor capsule (TABLE 5). There are several studies that have indicated that the number of vascular invasive foci in the latter tumor type is predictive of risk of tumor recurrence [85, 86].

TABLE 5: Classification scheme for follicular derived neoplasms

Criteria Follicular Adenoma Minimally invasive Carcinoma Encapsulated angio-invasive carcinoma Widely invasive carcinoma
Capsule Complete Complete Usually complete Incomplete or absent
Capsular invasion Absent Mandatory Usually resent Usually present
Vascular invasion Absent Absent Mandatory Usually present
Growth pattern Follicular Usually follicular Follicular, solid, trabecular Often solid or trabecular or insular
Anaplasia Minimal Minimal Minimal Variable



In examining a thyroid with an encapsulated tumor, it is critical to adequately sample the tumor and to thoroughly review the sections that demonstrate the capsule. It is usually beneficial to submit and examine the entire capsule; the central component of the tumor can be sampled minimally. Capsular and vascular invasion can be extremely focal and limited to one or two sections of the tumor. This is one reason that frozen section is fairly ineffectual at identifying malignancy in follicular patterned encapsulated tumors [87]. In one study, in fact, invasive foci were only seen in 1 out of every 9 sections [88].

The inclusion of the category of "encapsulated angio-invasive follicular carcinoma" has not been adopted everywhere, nor has it been widely reported in the literature. Many pathologists still lump tumors with capsular invasion alone with those with vascular invasion into the category of "minimally invasive follicular carcinoma". While this practice is still the standard in many places, it is not optimal. Many studies have now demonstrated that vascular invasion is an independent predictor of risk in follicular carcinomas [84, 85, 89, 90]. The value of counting the number of vascular invasive foci has been disputed, with some studies indicating a worse risk with more foci, and others not seeing this correlation [86, 91]. Extensive vascular invasion may also be accounted for if one uses the terminology of "poorly differentiated thyroid carcinoma", since many tumors with extensive vascular invasion will have other poor prognostic features, such as necrosis and trabecular/insular/solid growth patterns [92].

The histologic criteria that are used to define capsular are important, since they are the only indicators of malignancy in otherwise well-encapsulated tumors. The definitions have not been well established, and are somewhat controversial. Vascular invasive foci should show islands, nests, polyps, or protrusions of tumor cells into capsular veins (not capillaries). This is usually accompanied by a reaction from the host to the tumor, in the form of either endothelialization or fibrin caps on the tumor emboli. The tumor is usually adherent to at least one wall of the vessel; if the tumor is free-floating, care must be taken to exclude artifactual pushing in of tumor during surgery or processing (TABLE 6 and FIGURE 4). Some pitfalls exist in identifying vascular invasion. One is excluding changes in the capsule that are secondary to FNA artifacts. Another includes vascular reactive changes that are unrelated to tumor, including pseudoangiomatous changes and Kaposi's-like proliferations [93]. There are some very interesting studies that were done with 3-dimensional reconstructions that suggested that capsular invasion is all related, in fact, to vascular invasion that has completely occluded and destroyed the penetrating capsular vessels [85].

TABLE 6: The criteria used to define angio-invasion

Criteria Description in true angio-invasion
Location Vessels at the capsule or beyond the capsule (not within the tumor parenchyma)
Involvement of vessel Tumor plugs, polyps, or protrusions into the vascular channel
Host reaction Tumor is usually endothelialized (covered by endothelium) or has fibrin associated with it.
Attachment Tumor is often attached to wall, but this is not necessary to make the diagnosis as long as other features are present



The basic definition of capsular invasion is invasion of the tumor beyond the capsule of the tumor and into the surrounding thyroid gland parenchyma. There is controversy, however, in whether the invasion needs to include full capsular penetration or if this can be partial. Some authors have argued that the tumor must touch the thyroid parenchyma on the other side of the capsule [94, 95]. Others, however, have shown that some tumors that are clearly malignant show incomplete capsular invasion (though they may have separate vascular invasion) [91]. If only partial capsular invasion is seen, the pathologist should ensure that the tumor has been adequately sampled, and then consider obtaining multiple deeper levels into the blocks. The criteria for making the diagnosis of capsular invasion are shown in TABLE 6 and FIGURE 5.

TABLE 6: The criteria used to define capsular invasion
Criteria Description in true capsular invasion
Location At the level of the capsule, apparent from low power
Shape/morphology Usually takes the shape of a mushroom, with a small neck through the capsule and blossoming on the other side into the parenchyma
Level of invasion Controversy exists about whether tumor that is invading into the capsule, but not entirely through the capsule, reflects invasion.
Host reaction Usually capsular invasion does not have significant inflammation, hemosiderin, or reaction. These changes may bring up the possibility of artifactual trapping in the post-FNA setting.



One difficult histologic finding to resolve is the presence of multiple nodules of encapsulated tumor. This type of situation may arise with one dominant tumor nodule and a separate smaller focus that is present outside of and adjacent to the main mass. Are these capsular invasive foci in which the connection is no longer seen? The situation may also arise in the setting of multifocal, bilateral apparently encapsulated nodules. Is this evidence of widely invasive carcinoma? Are they clonally related or separate clones of tumor developing synchronously? In the first scenario, if the tumor section is from one pole of the tumor (i.e., the top slice or the bottom slice) the capsule will often be irregular. These foci can usually be disregarded as sectioning artifact that derives from taking a flat section from a round surface. If the sections are deeper within the lesion, however, they should be investigated with levels into that particular tumor block. In the second scenario, this can rarely be resolved at the microscopic level. They can be examined at the molecular level, however, and clonality can be confirmed.





Widely invasive follicular and Hürthle cell carcinomas are those that are either multifocal in the thyroid gland or have invasion beyond the thyroid gland [96]. Occasionally, encapsulated unifocal tumors will have such aggressive vascular invasion and other markers of poor differentiation (trabecular/insular/solid growth, necrosis, or mitotic activity) that they may behave like one of widely invasive tumors. Widely invasive carcinomas have an intermediate prognosis: that is, somewhere between minimally invasive or angioinvasive tumors and undifferentiated tumors. 5-year survivals are in the 50-60% range, similar in many ways to the tumors that have been classified as "poorly differentiated carcinomas" [97, 98]. There are no immunohistochemical stains that can aid in making the diagnosis of minimally, angio-invasive, or widely invasive follicular derived carcinomas. There are, however, some unique and interesting molecular mutations that have been described in these tumors.

Molecular Mutational Events in Follicular Lesions
The RAS genes (K-RAS, N-RAS, and H-RAS) have long been implicated in the pathogenesis of follicular derived tumors [99]. In most series, a RAS gene mutation is found in up to 40-50% of follicular derived tumors [100]. Recent evidence suggests that RAS gene mutations may be much more common in more aggressive follicular neoplasms and may in fact be a marker for aggressive behavior [101].

Up to 40% of FCCs are known to harbor a translocation between the PAX8 gene and the PPARγ gene (PAX8-PPARγ) [102]. The translocation is best identified at the mRNA level, but PPARγ protein over-expression has also been correlated with the translocation, though it is note entirely specific [103, 104]. Interestingly, when the group of tumors that have the translocation are compared to those that do not, several groups have described significant differences, both at the clinical level and at the molecular level; this suggests different pathways of development in PAX8-PPARγ tumors [100]. Translocations can be present in both adenomas and carcinomas, and malignant tumors with the translocation appear to have a better prognosis, be better differentiated and to have less frequent distant metastases [100, 103, 105]. Furthermore, in specific molecular and expression studies, the two groups separate out based on the presence or absence of the translocation [105, 106]. Hürthle cell carcinomas have not been found to harbor the PAX8-PPARγ translocation or high rates of RAS mutations [100, 107].

In studies of aneuploidy and loss of tumor suppressor genes in the follicular derived tumors, it has been found that the more aggressive the tumor is both clinically and histologically, the more frequent the rate of loss of heterozygosity [96, 107, 108, 109]. By using a panel of tumor suppressor genes (TSGs) in this type of analysis, high risk tumors can be identified by their mutational profile [96]. The follicular and Hurthle cell adenomas, which are considered to be the lowest grade of neoplasm, have low mutational burdens (mean fractional allelic loss [FAL]: 9%) The minimally invasive and angio-invasive carcinomas have marginally different mutational burdens and cannot be reliably separated from one another (mean FAL: 30%). The widely invasive follicular carcinomas, however, have high fractional allelic loss (mean FAL: 53%). These results indicate that the mutational burden correlates with our standard interpretations of the aggressiveness of tumors at the histologic level.

The panel of TSGs that was used was initially selected from the literature of all thyroid tumors that had been studied for loss of heterozygosity. These mutations were thought to be relatively early events in tumorigenesis, but not necessarily essential to the pathways of carcinogenesis for thyroid tumors. When assessing a panel of TSGs for loss and correlating with tumor grading, prognosis, or outcome, the mutational burden is used much more as a marker of genetic instability and aneuploidy than for specific events that contribute each individually to carcinogenesis [10]. Therefore, it is not anticipated that tumors will harbor similar mutational patterns or losses from one to another, but rather that the number of mutational events will correlate with the aggressiveness of the disease [110].

In a new study, we have further refined our panel of TSGs and performed this analysis of TSGs on a well characterized series of benign and malignant follicular derived tumors on which we have long-term follow-up data [111]. Again, the mutational burden, as measured with the FAL correlated extremely well with the histologic classification (Adenomas FAL = 14%, Minimally/angio-invasive FAL = 44%, and widely invasive FAL = 86%). Furthermore, tumors with recurrence or those that resulted in death from disease for the patient had a high rate of tumor suppressor gene loss. Tumors from patients that did not have evidence of recurrence had a mean FAL of 22%, while tumors that recurred or resulted in death from disease had a mean FAL of 78%.

One particularly interesting tumor that was from a patient who had recurrent disease; this tumor was classified histologically as an encapsulated angio-invasive tumor (not widely invasive. This tumor had a high mutational burden more similar to those seen in the widely invasive category. Our results indicate that high levels of tumor suppressor gene loss also correlate again with the histologic classification, but also strongly correlate with behavior of the tumor. In some interesting work examining the mitochondrial DNA, it has been found that HCCs do have mutations in these genes, though these mutations have also been seen in other types of thyroid cancer [112, 113].

The tumor suppressor gene profile can also be used to assess clonality in multifocal lesions. The panel of mutations is examined for each separate nodule and similar patterns of loss are seen this is evidence that the tumors are clonally related or presumably derived from the same precursor lesions.

References
  1. Knudson, A.G., Jr. and L.C. Strong, Mutation and cancer: a model for Wilms' tumor of the kidney. J Natl Cancer Inst, 1972. 48(2): p. 313-24.

  2. Knudson, A.G., Jr., Genetics and the etiology of childhood cancer. Pediatr Res, 1976. 10(5): p. 513-7.

  3. Payne, S.R. and C.J. Kemp, Tumor suppressor genetics. Carcinogenesis, 2005. 26(12): p. 2031-45.

  4. Iwasa, Y., et al., Population genetics of tumor suppressor genes. J Theor Biol, 2005. 233(1): p. 15-23.

  5. Fearon, E.R., S.R. Hamilton, and B. Vogelstein, Clonal analysis of human colorectal tumors. Science, 1987. 238(4824): p. 193-7.

  6. Vogelstein, B., et al., Genetic alterations during colorectal-tumor development. N Engl J Med, 1988. 319(9): p. 525-32.

  7. Thiagalingam, S., et al., Loss of heterozygosity as a predictor to map tumor suppressor genes in cancer: molecular basis of its occurrence. Current Opinion in Oncology, 2002. 14(1): p. 65-72.

  8. Duesberg, P., et al., Genetic instability of cancer cells is proportional to their degree of aneuploidy. Proc Natl Acad Sci U S A, 1998. 95(23): p. 13692-7.

  9. Boland, C.R. and A. Goel, Somatic evolution of cancer cells. Semin Cancer Biol, 2005. 15(6): p. 436-50.

  10. Duesberg, P., Does aneuploidy or mutation start cancer? Science, 2005. 307(5706): p. 41.

  11. Michor, F., et al., Can chromosomal instability initiate tumorigenesis? Semin Cancer Biol, 2005. 15(1): p. 43-9.

  12. Hunt, J.L. and S.D. Finkelstein, Microdissection techniques for molecular testing in surgical pathology. Arch Pathol Lab Med, 2004. 128(12): p. 1372-8.

  13. Roylance, R., Methods of molecular analysis: assessing losses and gains in tumours. Mol Pathol, 2002. 55(1): p. 25-8.

  14. Sakaki, Y., et al., Human versus chimpanzee chromosome-wide sequence comparison and its evolutionary implication. Cold Spring Harb Symp Quant Biol, 2003. 68: p. 455-60.

  15. Fujiyama, A., et al., Construction and analysis of a human-chimpanzee comparative clone map. Science, 2002. 295(5552): p. 131-4.

  16. Staiti, N., D. Di Martino, and L. Saravo, A novel approach in personal identification from tissue samples undergone different processes through STR typing. Forensic Sci Int, 2004. 146 Suppl: p. S171-3.

  17. Cohen, J., Genes and behavior make an appearance in the O. J. trial. Science, 1995. 268(5207): p. 22-3.

  18. Georgiou, M., A. Hatzaki, and A. Koutselinis, Identification of an alleged offender of murder by VNTR analysis: case report. Am J Forensic Med Pathol, 2000. 21(2): p. 162-5.

  19. Sweet, D.J. and C.H. Sweet, DNA analysis of dental pulp to link incinerated remains of homicide victim to crime scene. J Forensic Sci, 1995. 40(2): p. 310-4.

  20. Sanchez, J.J., et al., A multiplex assay with 52 single nucleotide polymorphisms for human identification. Electrophoresis, 2006. 27(9): p. 1713-1724.

  21. Grubwieser, P., et al., Evaluation of an extended set of 15 candidate STR loci for paternity and kinship analysis in an Austrian population sample. Int J Legal Med, 2006.

  22. Pretty, I.A. and D.P. Hildebrand, The forensic and investigative significance of reverse paternity testing with absent maternal sample. Am J Forensic Med Pathol, 2005. 26(4): p. 340-2.

  23. Hunt, J.L., et al., A microdissection and molecular genotyping assay to confirm the identity of tissue floaters in paraffin-embedded tissue blocks. Arch Pathol Lab Med, 2003. 127(2): p. 213-7.

  24. Tsongalis, G.J., et al., Applications of forensic identity testing in the clinical laboratory. American Journal of Clinical Pathology, 1999. 112(1 Suppl 1): p. S93-103.

  25. Slebos, R.J., et al., Analytical and statistical methods to evaluate microsatellite allelic imbalance in small amounts of DNA. Laboratory Investigation, 2004. 84(5): p. 649-57.

  26. Sieben, N.L., et al., PCR artifacts in LOH and MSI analysis of microdissected tumor cells.[comment]. Human Pathology., 2000. 31(11): p. 1414-9.

  27. Melton, L.J., 3rd, The epidemiology of primary hyperparathyroidism in North America. Journal of Bone & Mineral Research, 2002. 17(2).

  28. Adamietz, I.A., et al., [Prognostic factors and the effect of radiotherapy in the treatment of differentiated thyroid carcinomas]. Strahlentherapie und Onkologie, 1998. 174(12): p. 618-23.

  29. Locatelli, F., The need for better control of secondary hyperparathyroidism. Nephrology Dialysis Transplantation, 2004. 15(19).

  30. Vestergaard, P., et al., Cohort study of risk of fracture before and after surgery for primary hyperparathyroidism. Bmj, 2000. 321(7261): p. 598-602.

  31. Khosla, S. and J. Melton, 3rd, Fracture risk in primary hyperparathyroidism.[comment]. Journal of Bone & Mineral Research, 2002. 17(2).

  32. Lo, C.Y., et al., Surgical treatment for primary hyperparathyroidism in Hong Kong: changes in clinical pattern over 3 decades. Archives of Surgery, 2004. 139(1): p. 77-82.

  33. Palazzo, F.F., et al., Parathyroid autotransplantation during total thyroidectomy--does the number of glands transplanted affect outcome? World J Surg, 2005. 29(5): p. 629-31.

  34. Carty, S.E., Prevention and management of complications in parathyroid surgery. Otolaryngologic Clinics of North America, 2004. 37: p. 897-974.

  35. Kopp, I., et al., Predictive genetic screening and clinical findings in multiple endocrine neoplasia type I families. World Journal of Surgery., 2001. 25(5): p. 610-6.

  36. Mozes, G., et al., The predictive value of laboratory findings in patients with primary hyperparathyroidism. Journal of the American College of Surgeons, 2002. 194(2): p. 126-30.

  37. Ferzli, G., et al., Three new tools for parathyroid surgery: expensive and unnecessary?[see comment]. Journal of the American College of Surgeons, 2004. 198(3): p. 349-51.

  38. Clerici, T., et al., Impact of intraoperative parathyroid hormone monitoring on the prediction of multiglandular parathyroid disease. World Journal of Surgery, 2004. 28(2): p. 187-92.

  39. Carty, S.E., et al., The elevated parathormone level after concise parathyroid exploration for primary sporadic hyperparathyroidism. Surgery, 2002. 132: p. 1086-1093.

  40. Horanyi, J., et al., Parathyroid surgical failures with misleading falls of intraoperative parathyroid hormone levels. Journal of Endocrinological Investigation, 1095. 26(11): p. 1095-9.

  41. Bondeson, A.G., et al., Fat staining in parathyroid disease--diagnostic value and impact on surgical strategy: clinicopathologic analysis of 191 cases. Human Pathology, 1985. 16(12): p. 1255-63.

  42. Chen, K.T., Fat stain in hyperparathyroidism. American Journal of Surgical Pathology, 1982. 6(2): p. 191-2.

  43. Kasdon, E.J., et al., Surgical pathology of hyperparathyroidism. Usefulness of fat stain and problems in interpretation. American Journal of Surgical Pathology, 1981. 5(4): p. 381-4.

  44. Bergson, E.J. and K.S. Heller, The clinical significance and anatomic distribution of parathyroid double adenomas. Journal of the American College of Surgeons, 2004. 198(2): p. 185-9.

  45. Rosell, R., et al., Molecular markers and targeted therapy with novel agents: prospects in the treatment of non-small cell lung cancer. Lung Cancer, 2002. 4: p. 43-9.

  46. Yao, K., et al., Weight of normal parathyroid glands in patients with parathyroid adenomas. J Clin Endocrinol Metab, 2004. 89(7): p. 3208-13.

  47. Dufour, D.R. and S.Y. Wilkerson, Factors related to parathyroid weight in normal persons. Arch Pathol Lab Med, 1983. 107(4): p. 167-72.

  48. Kikuchi, S., et al., Complication of thyroidectomy in patients with radiation-induced thyroid neoplasms. Archives of Surgery, 2004. 139(11): p. 1185-8.

  49. Fleischer, J., et al., Oxyphil parathyroid adenoma: a malignant presentation of a benign disease. J Clin Endocrinol Metab, 2004. 89(12): p. 5948-51.

  50. Giorgadze, T., et al., Oncocytic parathyroid adenoma: problem in cytological diagnosis. Diagn Cytopathol, 2004. 31(4): p. 276-80.

  51. Moran, C.A. and S. Suster, Primary parathyroid tumors of the mediastinum: a clinicopathologic and immunohistochemical study of 17 cases. Am J Clin Pathol, 2005. 124(5): p. 749-54.

  52. Grenko, R.T., et al., Water-clear cell adenoma of the parathyroid. A case report with immunohistochemistry and electron microscopy. Arch Pathol Lab Med, 1995. 119(11): p. 1072-4.

  53. Kuhel, W.I., D.I. Kutler, and C.A. Santos-Buch, Poorly differentiated insular thyroid carcinoma. A case report with identification of intact insulae with fine needle aspiration biopsy. Acta Cytologica, 1998. 42(4): p. 991-7.

  54. Parfitt, A.M., Q. Wang, and S. Palnitkar, Rates of cell proliferation in adenomatous, suppressed, and normal parathyroid tissue: implications for pathogenesis. J Clin Endocrinol Metab, 1998. 83(3): p. 863-9.

  55. DeLellis, R.A., Parathyroid carcinoma: an overview. Adv Anat Pathol, 2005. 12(2): p. 53-61.

  56. Iacobone, M., F. Lumachi, and G. Favia, Up-to-date on parathyroid carcinoma: analysis of an experience of 19 cases. Journal of Surgical Oncology, 2004. 88(4): p. 223-8.

  57. Beus, K.S. and B.C. Stack, Jr., Parathyroid carcinoma. Otolaryngologic Clinics of North America, 2004. 37(4): p. 845-54.

  58. Rodgers, S.E. and N.D. Perrier, Parathyroid carcinoma. Curr Opin Oncol, 2006. 18(1): p. 16-22.

  59. Castleman, B., A. Schantz, and S. Roth, Parathyroid hyperplasia in primary hyperparathyroidism: a review of 85 cases. Cancer, 1976. 38(4): p. 1668-75.

  60. Stojadinovic, A., et al., Parathyroid neoplasms: clinical, histopathological, and tissue microarray-based molecular analysis. Human Pathology, 2003. 34(1): p. 54-64.

  61. Chevillard, S., et al., Gene expression profiling of differentiated thyroid neoplasms: diagnostic and clinical implications. Clinical Cancer Research, 2004. 10(19): p. 6586-97.

  62. Koea, J.B. and J.H. Shaw, Parathyroid cancer: biology and management. Surg Oncol, 1999. 8(3): p. 155-65.

  63. Streeten, E.A., et al., Studies in a kindred with parathyroid carcinoma. Journal of Clinical Endocrinology & Metabolism, 1992. 75(2): p. 362-6.

  64. Kleinpeter, K.P., et al., Is parathyroid carcinoma indeed a lethal disease? Ann Surg Oncol, 2005. 12(3): p. 260-6.

  65. Kameyama, K. and H. Takami, Proposal for the histological classification of parathyroid carcinoma. Endocr Pathol, 2005. 16(1): p. 49-52.

  66. Taniguchi, K., et al., Differentiation of follicular thyroid adenoma from carcinoma by means of gene expression profiling with adapter-tagged competitive polymerase chain reaction. Oncology, 2005. 69(5): p. 428-35.

  67. Evans, C.F., L. Mansfield, and A.K. Sharma, Recurrent hyperparathyroidism caused by parathyromatosis. Hosp Med, 2005. 66(7): p. 424-5.

  68. Lee, P.C., et al., Parathyromatosis: a cause for recurrent hyperparathyroidism. Endocr Pract, 2001. 7(3): p. 189-92.

  69. Villablanca, A., et al., Genetic and clinical characterization of sporadic cystic parathyroid tumours. Clinical Endocrinology, 2002. 56(2): p. 261-9.

  70. Valimaki, S., et al., Distinct target regions for chromosome 1p deletions in parathyroid adenomas and carcinomas. International Journal of Oncology, 2002. 21(4): p. 727-35.

  71. Rubio, M.P., et al., The putative glioma tumor suppressor gene on chromosome 19q maps between APOC2 and HRC. Cancer Research, 1994. 54(17): p. 4760-3.

  72. McPherson, J.D., et al., A physical map of the human genome. Nature., 2001. 409(6822): p. 934-41.

  73. Cryns, V.L., et al., Frequent loss of chromosome arm 1p DNA in parathyroid adenomas. Genes, Chromosomes & Cancer, 1995. 13(1): p. 9-17.

  74. Cetani, F., et al., A reappraisal of the Rb1 gene abnormalities in the diagnosis of parathyroid cancer. Clinical Endocrinology, 2004. 60(1): p. 99-106.

  75. Cetani, F., et al., Genetic analysis of the MEN1 gene and HPRT2 locus in two Italian kindreds with familial isolated hyperparathyroidism. Clinical Endocrinology, 2002. 56(4): p. 457-64.

  76. Vaidya, B., et al., Evidence for a new Graves disease susceptibility locus at chromosome 18q21. American Journal of Human Genetics., 2000. 66(5): p. 1710-4.

  77. Cryns, V.L., et al., Loss of the retinoblastoma tumor-suppressor gene in parathyroid carcinoma. New England Journal of Medicine, 1994. 330(11): p. 757-61.

  78. Shattuck, T.M., et al., Mutational analyses of RB and BRCA2 as candidate tumour suppressor genes in parathyroid carcinoma. Clinical Endocrinology, 2003. 59(2): p. 180-9.

  79. Hunt, J.L., et al., Allelic loss in parathyroid neoplasia can help characterize malignancy. Am J Surg Pathol, 2005. 29(8): p. 1049-55.

  80. Szijan, I., et al., Alterations in the retinoblastoma pathway of cell cycle control in parathyroid tumors. Oncology Reports, 2000. 7(2): p. 421-5.

  81. Shattuck, T.M., et al., Mutational analysis of Smad3, a candidate tumor suppressor implicated in TGF-beta and menin pathways, in parathyroid adenomas and enteropancreatic endocrine tumors. Journal of Clinical Endocrinology & Metabolism, 2002. 87(8): p. 3911-4.

  82. Awwad, R.A., et al., The role of transforming growth factor alpha in determining growth factor independence. Cancer Research, 2003. 63(15): p. 4731-8.

  83. Hunt, J.L., Unusual thyroid tumors: a review of pathologic and molecular diagnosis. Expert Review of Molecular Diagnostics, 2005. 5(5): p. 725-34.

  84. Thompson, L.D., et al., A clinicopathologic study of minimally invasive follicular carcinoma of the thyroid gland with a review of the English literature. Cancer, 2001. 91(3): p. 505-24.

  85. Aida, N., et al., 3-D analysis of vascular and capsular invasion in thyroid follicular carcinoma. Pathology International, 2001. 51(6): p. 425-30.

  86. Besic, N., M. Auersperg, and R. Golouh, Prognostic factors in follicular carcinoma of the thyroid--a multivariate survival analysis. European Journal of Surgical Oncology, 1999. 25(6): p. 599-605.

  87. Multanen, M., et al., The value of ultrasound-guided fine-needle aspiration biopsy (FNAB) and frozen section examination (FS) in the diagnosis of thyroid cancer. Annales Chirurgiae et Gynaecologiae, 1999. 88(2): p. 132-5.

  88. Leteurtre, E., et al., Why do frozen sections have limited value in encapsulated or minimally invasive follicular carcinoma of the thyroid? American Journal of Clinical Pathology, 2001. 115(3): p. 370-4.

  89. Nishida, T., S. Katayama, and M. Tsujimoto, The clinicopathological significance of histologic vascular invasion in differentiated thyroid carcinoma. American Journal of Surgery., 2002. 183(1): p. 80-6.

  90. Baloch, Z.W. and V.A. LiVolsi, Prognostic factors in well-differentiated follicular-derived carcinoma and medullary thyroid carcinoma. Thyroid., 2001. 11(7): p. 637-45.

  91. Goldstein, N.S., P. Czako, and J.S. Neill, Metastatic minimally invasive (encapsulated) follicular and Hurthle cell thyroid carcinoma: a study of 34 patients. Modern Pathology, 2000. 13(2): p. 123-30.

  92. Volante, M., et al., Poorly differentiated carcinomas of the thyroid with trabecular, insular, and solid patterns: a clinicopathologic study of 183 patients. Cancer, 2004. 100(5): p. 950-7.

  93. Baloch, Z.W. and V.A. LiVolsi, Intravascular Kaposi's-like spindle cell proliferation of the capsular vessels of follicular-derived thyroid carcinomas. Modern Pathology, 1998. 11(10): p. 995-8.

  94. Schmid, K.W. and N.R. Farid, How to define follicular thyroid carcinoma? Virchows Arch, 2006. 448(4): p. 385-393.

  95. Franssila, K.O., et al., Follicular carcinoma. Semin Diagn Pathol, 1985. 2(2): p. 101-22.

  96. Hunt, J.L., et al., A novel microdissection and genotyping of follicular-derived thyroid tumors to predict aggressiveness. Human Pathology, 2003. 34(4): p. 375-80.

  97. Collini, P., et al., Minimally invasive (encapsulated) follicular carcinoma of the thyroid gland is the low-risk counterpart of widely invasive follicular carcinoma but not of insular carcinoma. Virchows Archiv, 2003. 442(1): p. 71-6.

  98. Gillenwater, A.M. and R.S. Weber, Thyroid carcinoma. Cancer Treatment & Research, 1997. 90: p. 149-69.

  99. Capella, G., et al., Ras oncogene mutations in thyroid tumors: polymerase chain reaction-restriction-fragment-length polymorphism analysis from paraffin-embedded tissues. Diagnostic Molecular Pathology, 1996. 5(1): p. 45-52.

  100. Nikiforova, M.N., et al., RAS point mutations and PAX8-PPAR gamma rearrangement in thyroid tumors: evidence for distinct molecular pathways in thyroid follicular carcinoma. Journal of Clinical Endocrinology & Metabolism, 2003. 88(5): p. 2318-26.

  101. Garcia-Rostan, G., et al., ras mutations are associated with aggressive tumor phenotypes and poor prognosis in thyroid cancer. Journal of Clinical Oncology, 2003. 21(17): p. 3226-35.

  102. Kroll, T.G., et al., PAX8-PPARgamma1 fusion oncogene in human thyroid carcinoma [corrected]. Science, 2000. 289(5483): p. 1357-60.

  103. Cheung, L., et al., Detection of the PAX8-PPAR gamma fusion oncogene in both follicular thyroid carcinomas and adenomas. Journal of Clinical Endocrinology & Metabolism, 2003. 88(1): p. 354-7.

  104. Nakabashi, C.C., et al., The expression of PAX8-PPARgamma rearrangements is not specific to follicular thyroid carcinoma. Clinical Endocrinology, 2004. 61(2): p. 280-2.

  105. French, C.A., et al., Genetic and biological subgroups of low-stage follicular thyroid cancer. American Journal of Pathology, 2003. 162(4): p. 1053-60.

  106. Lui, W.O., et al., Expression profiling reveals a distinct transcription signature in follicular thyroid carcinomas with a PAX8-PPARgamma fusion oncogene. Oncogene, 2004. December 20: p. 1-10.

  107. Tallini, G., et al., Frequent chromosomal DNA unbalance in thyroid oncocytic (Hurthle cell) neoplasms detected by comparative genomic hybridization. Laboratory Investigation, 1999. 79(5): p. 547-55.

  108. Stoler, D.L., et al., Genomic instability measurement in the diagnosis of thyroid neoplasms. Head & Neck, 2002. 24(3): p. 290-5.

  109. Erickson, L.A., et al., Analysis of Hurthle cell neoplasms of the thyroid by interphase fluorescence in situ hybridization. American Journal of Surgical Pathology, 2001. 25(7): p. 911-7.

  110. Duesberg, P., A. Fabarius, and R. Hehlmann, Aneuploidy, the primary cause of the multilateral genomic instability of neoplastic and preneoplastic cells. IUBMB Life, 2004. 56(2): p. 65-81.

  111. Hunt, J.L., J.H. Yim, and S.E. Carty, Tumor suppressor gene loss predicts prognosis in follicular derived carcinomas. Thyroid, 2006. In Press.

  112. Tong, B.C., et al., Mitochondrial DNA alterations in thyroid cancer. Journal of Surgical Oncology, 2003. 82(3): p. 170-3.

  113. Maximo, V. and M. Sobrinho-Simoes, Mitochondrial DNA 'common' deletion in Hurthle cell lesions of the thyroid. Journal of Pathology, 2000. 192(4): p. 561-2.