PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Ann Surg Oncol. Author manuscript; available in PMC Sep 24, 2012.
Published in final edited form as:
PMCID: PMC3449317
NIHMSID: NIHMS407679
Identification of differentially expressed microRNA in parathyroid tumors
Reza Rahbari, MD,1 Alisha K. Holloway, PhD,2 Mei He, MD,1 Elham Khanafshar, MD,3 Orlo H. Clark, MD,4 and Electron Kebebew, MD1
1Endocrine Oncology Section, Surgery Branch, National Cancer Institute, Bethesda, Maryland
2Gladstone Institute University of California San Francisco, San Francisco, California
3Department of Pathology University of California San Francisco, San Francisco, California
4Department of Surgery, University of California San Francisco, San Francisco, California
Please address correspondence to: Electron Kebebew, MD National Cancer Institute Surgery Branch CRC, Room 4-5952 10 Center Drive, MSC 1201 Bethesda, MD 20892-1201 Phone: (301) 496-5049 Fax: (301) 402-1788 ; kebebewe/at/mail.nih.gov
Background
The molecular factors that control parathyroid tumorigenesis are poorly understood. In the absence of local invasion or metastasis, distinguishing benign from malignant parathyroid neoplasm is difficult on histologic examination. We studied the miRNA profile in normal, hyperplastic, and benign and malignant parathyroid tumors to better understand the molecular factors that may play a role in parathyroid tumorigenesis and that may serve as diagnostic markers for parathyroid carcinoma.
Methods
MiRNA arrays containing 825 human microRNAs with 4 duplicate probes per miRNA were used to profile parathyroid tumor (12 adenomas, 9 carcinomas, and 15 hyperplastic) samples normalized to 4 reference normal parathyroid glands. Differentially expressed miRNA were validated by real-time quantitative TaqMan PCR.
Results
One hundred and fifty six miRNAs in parathyroid hyperplasia, 277 microRNAs in parathyroid adenoma, and 167 microRNAs in parathyroid carcinomas were significantly dysregulated as compared to normal parathyroid glands (FDR < 0.05). By supervised clustering analysis, all parathyroid carcinomas clustered together. Three miRNAs (miR-26b, miR-30b and miR-126*) were significantly dysregulated between parathyroid carcinoma and parathyroid adenoma. Receiver operative characteristic curve analysis showed mir-126* was the best diagnostic marker, with an area under the curve of 0.776. MiRNAs are differentially expressed in parathyroid neoplasms.
Conclusions
Most miRNA are downregulated in parathyroid carcinoma while in parathyroid hyperplasia most miRNA are upregulated. MiRNA profiling shows distinct differentially expressed miRNAs by tumor type which may serve as helpful adjunct to distinguish parathyroid adenoma from carcinoma.
Keywords: Parathyroid Carcinoma, MicroRNA, Parathyroid Neoplasm, Hyperparathyroidism, Parathyroid Adenoma, Parathyroid Hyperplasia, Parathyroid tumors
Hyperparathyroidism is characterized by hypersecretion of parathyroid hormone. Hyperparathyroidism may be primary, secondary or rarely tertiary. Primary hyperparathyroidism occurs as a result of inappropriate parathyroid hormone secretion from enlarged parathyroid gland(s). Primary hyperparathyroidism is caused by a parathyroid adenoma (85%), parathyroid hyperplasia involving 4 glands (10%), double parathyroid adenomas (2–5%), or rarely due to parathyroid carcinoma (<1%)1. Secondary hyperparathyroidism occurs as a response usually due to low serum calcium levels associated with renal failure. In tertiary hyperparathyroidism, once the secondary cause of hyperparathyroidism is resolved there is continued autonomous hypersecretion of parathyroid hormone from an enlarged parathyroid gland.
Most parathyroid tumors in primary hyperparathyroidism are sporadic but approximately 5% are associated with autosomal dominant familial syndromes: Multiple Endocrine Neoplasia types 1 (MEN 1; OMIM #131100) and 2A (MEN 2A; OMIM #171400), Familial Isolated Hyperparathyroidism (OMIM #145000), and Hyperparathyroidism-Jaw Tumor Syndrome (HPT-JS; OMIM #145001)24. Parathyroid tumors are heterogeneous and histologic classification of tumors is often difficult. The pathologic distinction of parathyroid adenoma from carcinoma is also difficult in the absence of obvious local invasion and distant metastasis or disease recurrence5, 6. Moreover, if not grossly enlarged, parathyroid glands can also be difficult to histologically classify as hyperplastic3.
Several responsible germline genetic changes associated with primary hyperparathyroidism in familial syndrome have been identified (menin in MEN 1, RET in MEN 2A, CDC73/HRPT2 in HPT-JS) but most cases of hyperparathyroidism are sporadic2, 7. Several genetic changes have been implicated in a subset of sporadic parathyroid tumors. A chromosomal rearrangement of the cyclinD1 gene to the parathyroid hormone gene locus occurs, and cyclin D1 is overexpressed in up to 40% of sporadic parathyroid adenomas8. Recently, mutation in the tumor suppressor gene HRPT2 has also been identified in sporadic parathyroid carcinoma and a small subset of parathyroid adenomas6. Both the calcium sensing receptor (CaSR) and vitamin D receptor (VDR) may also play a role in parathyroid tumorigenesis4, 9.
MicroRNAs (miRNA, miR) are short, 19–22 nucleotides, non-coding RNAs. They account for 1% of the genome, and play a role in cellular processes such as apoptosis, proliferation and differentiation10, 11. MiRNAs are conserved across species and their expression is highly specific for tissue type. MiRNA regulate gene expression through mRNA degradation, translational modulation, and or gene silencing10, 12. Approximately, 30% of the genome is regulated by miRNA. In general, miRNAs are downregulated in most carcinoma and can function as either tumor suppressor or oncogene10, 12.
MiRNA profiling in several human malignancies have shown that such an approach may identify miRNAs with a role in tumor cell biology, to classify tumor subtypes, and to identify diagnostic and prognostic markers10. To further understand the molecular mechanisms involved in parathyroid tumorigenesis and, thus, improve clinical diagnosis of patients with primary hyerparathyroidism, we performed miRNA gene expression profiling in 40 parathyroid tumor samples (9 parathyroid carcinomas, 12 parathyroid adenomas, 15 parathyroid hyperplasia, with 4 reference normal parathyroid glands).
Patients and Parathyroid tissue samples
Parathyroid tissue samples including clinical and histopathologic data were obtained for 40 patients with approval of the Committee on Human Research at the University of California, San Francisco. Nine parathyroid carcinoma, 12 parathyroid adenoma and 15 parathyroid hyperplasia were obtained from 40 patients who had primary hyperparathyroidism. The 4 normal parathyroid gland samples were obtained from biopsy specimens at the time of neck exploration for parathyroidectomy. Cases of parathyroid carcinoma had Schantz and Castleman's histologic criteria and all cases had local invasion, recurrence and or distant metastasis6.
RNA extraction and Microarray preparation
Total RNA was extracted from fresh frozen tissue. At the time that tumor samples were sectioned for RNA extraction, representative portions of the tissue were examined by H & E histology. The quality of total RNA was determined with the Agilent 2100 Bioanalyzer and all samples had a RNA integrity number ≥ 7.0. MiRNA microarray profiling was done using the miRCURY LNA array version 11.0 (Exiqon). This array contains 7,720 probes, 3,300 of which represent 825 human miRNAs with 4 duplicate probes per miRNA. One μg of total RNA for each sample and pooled normal reference were labeled with Hy3 and Hy5 fluorescent label, respectively, using the miRCURY LNA Array power labeling kit (Exiqon), as described by the manufacturer. The Hy3-labeled samples and an Hy5-labeled reference RNA sample were mixed pair wise and hybridized to the miRCURY LNA array. The miRCURY LNA array microarray slides were scanned using the Agilent G2565BA Microarray Scanner System (Agilent Technologies, Inc.), and the image analysis was carried out using the ImaGene 7.0 software (BioDiscovery, Inc.).
Quantitative reverse transcriptase-PCR
MiRNAs were validated using quantitative reverse transcriptase-PCR. An adjusted p-value controlling for false discovery rate (FDR) <0.01 was used to identify miRNAs that were differentially expressed between adenoma and carcinoma. Thirteen out of the 24 miRNA were commercially available. Quantitative reverse transcriptase-PCR was used to quantify miRNA levels [hsa-miR-26b (assay# 000407), hsa-miR-27a (assay#000408), hsa-miR-27b (assay# 000409), hsa-miR 30b (assay # 000602), hsa-miR-28 (assay# 000411), hsa-miR-34a (assay# 000426), hsa-miR-100 (assay# 000437), hsa-miR-126 (assay# 000450), hsa-miR-126* (assay# 000451), hsa-miR-145 (assay# 000467), hsa-miR-423-3P (assay# 002626), hsa-let-7a (assay # 000377) and hsa-let-7f (assay# 000382)] using primers and probe purchased from Applied Biosystem (Foster City, CA). The gene expression pattern of these tumors was compared with a reference sample that consisted of a pooled sample of four normal parathyroid glands. Quantitative reverse transcriptase-PCRs (RT-PCR) for each sample were done in triplicate. Total RNA (10 ng for 10μl reaction) was converted to cDNA using primers specific for each miRNA and the TaqMan MiRNA reverse transcription kit (4366597; Applied Biosystems), according to the manufacturer's instructions. All quantitative reverse transcriptase-PCRs were done using a 5′ nuclease technique with specific Taqman Gene Expression Assays (Applied Biosystems) and Taqman Universal PCR Master Mix, NO AmpErase UNG (Applied Biosystems), on an ABI 7900HT Fast Real-Time PCR System. MiRNA expression level was expressed as the difference (ΔCt) between cycle threshold (Ct) for the miRNA of interest and pooled normal samples.
Target analysis
We conducted a literature search and identified genes of interest in parathyroid carcinoma. The genes of interest were CDC73 (HRPT2, parafibromin), Rb1, Galectin 3 (Lgals 3), Ki-67 (MKI67), CCND 2, CASR or VDR3, 4, 8, 9. We then used the miR-Ontology data base from Ferro Lab Data Mining and Informatics Group website that cross references Target Scan, miRanda and PicTar data base to search for potential targets13.
Statistical analysis
The microarray GPR files were loaded into R/Bioconductor using the marray package (Gentleman 2004). Flagged spots were removed from subsequent analysis and the remaining probes were used for normalization and subsequent analyses. The log2 ratio of the intensity of Cy5 to Cy3 signals were calculated for each miRNA on every array (with no background subtraction) and normalized by print tip loess normalization14, 15. Since individual miRNAs were represented by more than one probe on the array, the median of normalized log2 ratio of the replicate probes (for those with more than one unflagged probe) was used as the value for the miRNA. The summarized log2 ratios for each experiment were then used in moderated t-statistics and p-value calculation using the limma package in R/Bioconductor16, 17 with adjustment for false discovery rate using the Benjamini-Hochberg method. A linear model with one factor, tumor type, with three groups (Adenoma, Carcinoma and Hyperplasia) was established, followed by linear contrasts (Adenoma vs. Carcinoma, Adenoma vs. Hyperplasia, and Carcinoma vs. Hyperplasia). Adjusted p-value controlling for FDR using the Benjamini-Hochberg method were calculated. Unsupervised clustering analysis was performed using hierarchical clustering of the top (50,100 and 200) most variably expressed miRNAs based on the median absolute deviation using the Euclidean distance as a similarity metric. Supervised clustering analysis using the random Forest method in Bioconductor was performed using miRNAs that were uniquely differentially expressed between Adenoma and Carcinoma. The quantitative reverse transcriptase-PCR data was analyzed by using the XLSTAT statistical software. Pearson correlation was used to compare the array data with the real-time PCR data with r>0.65 showing strong correlation. The Mann-Whitney U test was used to compare non-parametric data between tumor types. Receiver operative characteristic curves were constructed using the Wald two-tailed 95% confidence intervals with the Bamber continuity correction.
MiRNA Cluster Analysis
We analyzed miRNA expression in 9 parathyroid carcinoma, 12 parathyroid adenoma and 15 parathyroid hyperplasia normalized to 4 reference pooled normal parathyroid glands. Unsupervised cluster analysis using the top 50 most variable miRNAs as defined by the median absolute deviation showed two main groups with modest clustering by tumor type Figure 1. Supervised cluster analysis using the random Forest method for differentially expressed miRNA that were uniquely dysregulated in adenoma versus carcinoma (FDR<0.05) showed complete clustering by tumor type Figure 2.
Figure 1
Figure 1
Unsupervised cluster analysis of top 50 most variably expressed miRNAs as defined by median absolute deviation
Figure 2
Figure 2
Supervised cluster analysis using differentially expressed miRNA between adenoma and carcinoma with FDR <0.05
Differentially expressed miRNAs
We found that 277 miRNA were differentially expressed in parathyroid adenoma, 167 in parathyroid carcinoma and 156 in parathyroid hyperplasia as compared to normal (FDR<0.05) Table 1. We found some overlap in the differentially expressed miRNAs between the three tumor types as compared to pooled normal parathyroid samples but 50 miRNAs were unique to parathyroid carcinoma, 122 to parathyroid adenoma and 22 to parathyroid hyperplasia Figure 3A. The percentage of downregulated miRNAs unique to the three tumor types were significantly higher for parathyroid carcinoma (58%) and parathyroid adenoma (50%) than for parathyroid hyperplasia (27%) (p <0.04). We also found in parathyroid carcinoma there was greater fold change in the differentially expressed miRNAs. In parathyroid carcinoma, 11 miRNAs (22.0%) were downregulated by greater than 2-folds and 3 miRNAs (6.0%) were upregulated by greater than 2 folds. While for miRNAs dysregulated in parathyroid adenoma and parathyroid hyperplasia, only 1 miRNA for parathyroid adenoma was dysregulated by greater than 2 fold Table 1.
Table 1
Table 1
Differentially expressed miRNA by tumor type with FDR <0.05
Figure 3
Figure 3
A) Venn diagram of differentially expressed miRNA as compared to normal by tumor type using a FDR <0.05, B) Venn Diagram of differentially expressed miRNA as compared between tumor types using a FDR of < 0.1
Because distinguishing between parathyroid adenoma and carcinoma is difficult by histology, a comparison of miRNA expression differences between these groups was also done. We identified 91 miRNA that were differentially expressed between adenoma and carcinoma Figure 3B. The majority of these miRNA were downregulated (58.2%) with nearly a fourth of the miRNA being downregulated by greater than 2 fold Table 2.
Table 2
Table 2
Correlation of miRNA expression between real-time quantitative TaqMan PCR and miRNA array for 13 differentially expressed miRNAs and fold difference
Validation of miRNA expression profile by TaqMan RT PCR
Using a FDR<0.01 from our microarray data we identified 24 differentially expressed miRNA between adenoma versus carcinoma. Thirteen miRNAs were used to validate the miRNA array expression data by real-time quantitative PCR. For 3 out of 13 miRNA there was strong correlation with the array data (Table 2). Similar to our microarray data all 13 selected miRNA were downregulated in carcinoma compared to adenoma. Three of the 13 selected miRNAs were significantly downregulated between adenoma and carcinoma (miR-126*, miR-26b, miR-30b, P <0.05) (Table 2). The area under the receiver operative characteristic curve was determined for the 13 differentially expressed miRNA. The miRNA with the highest accuracy was miR-126* (AUC=0.766).
Target analysis for differentially expressed miRNAs
We conducted a comprehensive target search for the 24 differentially expressed miRNA between parathyroid adenoma and carcinoma that were used in the validation process. The search included genes identified through literature search of genes involved in parathyroid tumorigenesis. The genes of interest were CDC73/HRPT2, Rb1, Galectin 3 (Lgals 3), Ki-67 (MKI67), CCND 2, CASR and VDR3, 4, 8, 9. We used the miR-Ontology data base that cross references Target Scan, miRanda and PicTar data base to search for potential targets13. In our search we were able to identify the CDC73 gene as a predicted target for miR-28. We further looked at the relationship between CDC73 expression and miR-28 using quantitative RT-PCR and immunohistochemistry for CDC73 protein expression3. We found no correlation between miR-28 expression and CDC73 expression in the parathyroid carcinoma samples (data not shown).
In this study, we sought to determine if parathyroid neoplasm had distinct miRNA signature. We found that the miRNA expression profile by unsupervised cluster analysis showed clustering of most parathyroid carcinomas together. As expected, in the supervised cluster analysis there was complete segregation of parathyroid samples by tumor type (parathyroid adenoma versus parathyroid carcinoma). We further validated our results using RT-PCR. Approximately two-thirds (9 of 13) of the miRNA expression levels showed good correlation with the microarray data. All the validated miRNA had the same pattern of downregulation in carcinoma compared to adenoma. Furthermore, miR-126*, miR-26b and miR-30b were significantly differentially expressed between parathyroid adenoma and parathyroid carcinoma by quantitative RT-PCR. MiR-126* levels was the most accurate for differentiating between parathyroid carcinoma and parathyroid adenoma (71% sensitivity, 82% specificity).
Parathyroid carcinoma is rare and little is known about the genetic basis for it. Moreover, distinguishing between benign and malignant parathyroid tumors can be extremely difficult3, 5, 6, 18. It is not rare for a patient with presumed benign disease to be diagnosed with parathyroid carcinoma after developing local recurrence or distant metastases19. Some histologic criteria for identifying parathyroid carcinoma (thick capsule, fibrous bands, nuclear pleomorphism, increased mitosis, capsular invasion and vascular invasion) are subjective and leave room for misdiagnosis5. In fact, most parathyroid carcinomas are diagnosed after recurrent disease develops. Some markers (galectin-3 and Rb overexpression, loss of CDC73/HRPT2) may be helpful for diagnosing parathyroid carcinoma. The absence of parafibromin (CDC73/HRPT2) staining has been reported to have 31% to 100% sensitivity and 41% to 100% specificity for distinguishing benign from malignant parathyroid tumors3, 20, 21. The identification and use of adjunct markers to improve the accuracy of parathyroid carcinoma diagnosis may have several clinical implications. For example, those patients found to have histologic equivocal criteria for parathyroid carcinoma but suggestive of parathyroid carcinoma on marker analysis, for carcinoma, may have more close follow up or reoperation with more aggressive surgical resection if tumor margins are positive.
In recent years, there has been a great deal of advancement in our knowledge regarding the role of miRNA in tumorigenesis10. However, in the current literature there are no studies that have attempted to show the miRNA signature for parathyroid carcinoma, adenoma and hyperplasia. There has been one study which has evaluated miRNA expression profiling in a limited number of samples and parathyroid tumor types. Corbetta and colleagues found 17 miRNA differential expressed between 4 parathyroid carcinoma samples compared to 2 normal parathyroid glands out of 362 miRNAs profiled22. In contrast to our findings, unsupervised hierarchical cluster analysis did not show separate clustering of parathyroid tumor types comparing the 6 samples. The investigators validated 4 miRNAs (miR-139, miR-296, miR-222 and miR-503) to be differentially expressed in 4 parathyroid carcinoma samples compared to 2 normal parathyroid gland samples. None of the miRNAs identified to be differentially expressed in their study were differentially expressed in our study. Our study design and array platform used was different than the study by Corbetta et al. as was the goal of our study22. We profiled a greater number of miRNAs among a greater number of parathyroid tumor types using a larger number of samples per histologic group, which were compared to reference pooled normal parathyroid glands. The strength of such an approach is it allowed us to determine if the miRNA profile was different among the different tumor types providing at least some clues to the miRNA signature of parathyroid tumor types, for what often can be a clinically and histologically heterogeneous group of tumor types. Thus, the clustering of most parathyroid carcinoma samples together by unsupervised analysis lends some support to a miRNA signature being present which is emphasized with the complete separation between parathyroid carcinoma and adenoma by supervised cluster analysis.
A formal diagnostic accuracy analysis of the significantly differentially expressed miRNAs between parathyroid adenoma and carcinoma showed one miRNA with good accuracy (miR-126*) by quantitative RT PCR. For these reasons, we believe are study results are unique and provide some insight into miRNA expression in parathyroid tumors and identify possible candidate miRNAs which may have a role in parathyroid tumorigenesis.
Again, by unsupervised clustering all but one parathyroid carcinoma grouped in two distinct clusters. For parathyroid hyperplasia and parathyroid adenoma there was no distinct pattern observed in the unsupervised clustering. One parathyroid adenoma and one parathyroid hyperplasia clustered with parathyroid carcinoma. Given the current limitations in the clinical and histologic classification of some parathyroid tumors, it is unclear if these samples were misclassified as parathyroid adenoma and hyperplasia but only long term clinical follow up will be able to help clarify this issue.
We also conducted a target search for the 13 differentially expressed miRNA between carcinoma and adenoma (FDR<0.01) looking at genes implicated in parathyroid dysregulation as potential targets. The genes of interest were CDC73/HRPT2, Rb1, Galectin 3 (Lgals 3), Ki-67 (MKI67), CCND 2, CASR or VDR3, 4, 8, 9. Our search identified CDC73 as a potential target for miR-28, but we did not find a relationship between miR-28 and CDC73 levels using RT PCR and immunohistochemistry. We also examined the chromosomal location of the 13 miRNAs and none of the target genes and miRNAs shared a common chromosomal location. The most common reported chromosomal imbalance in parathyroid carcinoma is loss of the 1q21-q32 region which is the coding region for CDC 737. None of our miRNAs were located in that region. Another common site for chromosomal imbalance is at 9q33-qter, this imbalance is reported in up to 24% of tumor samples23. This region is the same chromosomal region for miR-126* (9q 34.3). Kytola and colleagues reported this region to be a region of genomic gain by comparative genomic hybridization but miR-126* (the minor strand) and miR-126 (the major strand) were both downregulated in our carcinoma samples 23.
The 3 miRNAs (miR-126*, miR-30b, miR-26b) that were significantly dysregulated in carcinoma versus adenoma have been implicated in several malignancies2429. For example, the major strand of miR-126* has been implicated in non-small cell carcinoma30, 31, progression of adenoma to carcinoma in colon cancer32 and some leukemia33. Mir-126 is believed to regulate the epidermal growth factor-like domain 7 (EGFL7), which is involved in cellular migration and angiogenesis31. Also, an upregulation in miR-126 has been shown in vitro to decrease growth rate of cell lines from lung cancer origin30.
In summary, miRNAs are differentially expressed in parathyroid neoplasms. Most miRNA are downregulated in parathyroid carcinoma while in parathyroid hyperplasia most miRNA are upregulated. MiRNA profiling shows distinct differentially expressed miRNAs by tumor type which may serve as helpful adjunct to distinguish parathyroid adenoma from carcinoma.
Synopsis
A large number of miRNA are downregulated in parathyroid carcinoma while in parathyroid hyperplasia miRNA are upregulated. MiRNA profiling shows distinct differentially expressed miRNAs by tumor type which may serve as helpful adjunct to distinguish parathyroid adenoma from carcinoma.
Acknowledgements
None
Funding: This work was supported in part by grants from the UCSF Comprehensive Cancer Center, American Cancer Society, and the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.
Footnotes
Declaration of interest: There is no conflict of interest.
1. Kebebew E, Clark OH. Parathyroid adenoma, hyperplasia, and carcinoma: localization, technical details of primary neck exploration, and treatment of hypercalcemic crisis. Surg Oncol Clin N Am. 1998;7(4):721–48. [PubMed]
2. Carpten JD, Robbins CM, Villablanca A, et al. HRPT2, encoding parafibromin, is mutated in hyperparathyroidism-jaw tumor syndrome. Nat Genet. 2002;32(4):676–80. [PubMed]
3. Fernandez-Ranvier GG, Khanafshar E, Tacha D, et al. Defining a molecular phenotype for benign and malignant parathyroid tumors. Cancer. 2009;115(2):334–44. [PubMed]
4. Haven CJ, Howell VM, Eilers PH, et al. Gene expression of parathyroid tumors: molecular subclassification and identification of the potential malignant phenotype. Cancer Res. 2004;64(20):7405–11. [PubMed]
5. Fujimoto Y, Obara T. How to recognize and treat parathyroid carcinoma. Surg Clin North Am. 1987;67(2):343–57. [PubMed]
6. Schantz A, Castleman B. Parathyroid carcinoma. A study of 70 cases. Cancer. 1973;31(3):600–5. [PubMed]
7. Shattuck TM, Valimaki S, Obara T, et al. Somatic and germ-line mutations of the HRPT2 gene in sporadic parathyroid carcinoma. N Engl J Med. 2003;349(18):1722–9. [PubMed]
8. Mallya SM, Arnold A. Cyclin D1 in parathyroid disease. Front Biosci. 2000;5:D367–71. [PubMed]
9. Yano S, Sugimoto T, Tsukamoto T, et al. Decrease in vitamin D receptor and calcium-sensing receptor in highly proliferative parathyroid adenomas. Eur J Endocrinol. 2003;148(4):403–11. [PubMed]
10. Iorio MV, Croce CM. MicroRNAs in cancer: small molecules with a huge impact. J Clin Oncol. 2009;27(34):5848–56. [PMC free article] [PubMed]
11. Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell. 1993;75(5):843–54. [PubMed]
12. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell. 2004;116(2):281–97. [PubMed]
13. Lagana A, Forte S, Giudice A, et al. miRo: a miRNA knowledge base. Database (Oxford) 2009;2009:bap008. [PMC free article] [PubMed]
14. Smyth GK, Speed T. Normalization of cDNA microarray data. Methods. 2003;31(4):265–73. [PubMed]
15. Yang YH, Dudoit S, Luu P, et al. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res. 2002;30(4):e15. [PMC free article] [PubMed]
16. Smyth G. limma: Linear Models for Microarray Data. In: Robert Gentleman VJC WH, Irizarry Rafael A., Dudoit Sandrine., editors. Bioinformatics and Computational Biology Solutions Using R and Bioconductor. Springer; New York: 2005. pp. 397–420.
17. Smyth GK. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol. 2004;3(Article3) [PubMed]
18. Lee PK, Jarosek SL, Virnig BA, et al. Trends in the incidence and treatment of parathyroid cancer in the United States. Cancer. 2007;109(9):1736–41. [PubMed]
19. Kebebew E, Arici C, Duh QY, Clark OH. Localization and reoperation results for persistent and recurrent parathyroid carcinoma. Arch Surg. 2001;136(8):878–85. [PubMed]
20. Howell VM, Gill A, Clarkson A, et al. Accuracy of combined protein gene product 9.5 and parafibromin markers for immunohistochemical diagnosis of parathyroid carcinoma. J Clin Endocrinol Metab. 2009;94(2):434–41. [PubMed]
21. Cetani F, Ambrogini E, Viacava P, et al. Should parafibromin staining replace HRTP2 gene analysis as an additional tool for histologic diagnosis of parathyroid carcinoma? Eur J Endocrinol. 2007;156(5):547–54. [PubMed]
22. Corbetta S, Vaira V, Guarnieri V, et al. Differential expression of microRNAs in human parathyroid carcinomas compared with normal parathyroid tissue. Endocr Relat Cancer. 17(1):135–146. [PubMed]
23. Kytola S, Farnebo F, Obara T, et al. Patterns of chromosomal imbalances in parathyroid carcinomas. Am J Pathol. 2000;157(2):579–86. [PubMed]
24. Lu Y, Ryan SL, Elliott DJ, et al. Amplification and overexpression of Hsa-miR-30b, Hsa-miR-30d and KHDRBS3 at 8q24.22-q24.23 in medulloblastoma. PLoS One. 2009;4(7):e6159. [PMC free article] [PubMed]
25. Guled M, Lahti L, Lindholm PM, et al. CDKN2A, NF2, and JUN are dysregulated among other genes by miRNAs in malignant mesothelioma -A miRNA microarray analysis. Genes Chromosomes Cancer. 2009;48(7):615–23. [PubMed]
26. Jardin F, Callanan M, Penther D, et al. Recurrent genomic aberrations combined with deletions of various tumour suppressor genes may deregulate the G1/S transition in CD4+CD56+ haematodermic neoplasms and contribute to the aggressiveness of the disease. Leukemia. 2009;23(4):698–707. [PubMed]
27. Zanette DL, Rivadavia F, Molfetta GA, et al. miRNA expression profiles in chronic lymphocytic and acute lymphocytic leukemia. Braz J Med Biol Res. 2007;40(11):1435–40. [PubMed]
28. Ji J, Shi J, Budhu A, et al. MicroRNA expression, survival, and response to interferon in liver cancer. N Engl J Med. 2009;361(15):1437–47. [PMC free article] [PubMed]
29. Liu X, Chen Z, Yu J, et al. MicroRNA Profiling and Head and Neck Cancer. Comp Funct Genomics. 2009:837514. [PMC free article] [PubMed]
30. Liu B, Peng XC, Zheng XL, et al. MiR-126 restoration down-regulate VEGF and inhibit the growth of lung cancer cell lines in vitro and in vivo. Lung Cancer. 2009;66(2):169–75. [PubMed]
31. Sun Y, Bai Y, Zhang F, et al. miR-126 inhibits non-small cell lung cancer cells proliferation by targeting EGFL7. Biochem Biophys Res Commun. 391(3):1483–9. [PubMed]
32. Guo C, Sah JF, Beard L, et al. The noncoding RNA, miR-126, suppresses the growth of neoplastic cells by targeting phosphatidylinositol 3-kinase signaling and is frequently lost in colon cancers. Genes Chromosomes Cancer. 2008;47(11):939–46. [PMC free article] [PubMed]
33. Hussein K, Dralle W, Theophile K, et al. Megakaryocytic expression of miRNA 10a, 17-5p, 20a and 126 in Philadelphia chromosome-negative myeloproliferative neoplasm. Ann Hematol. 2009;88(4):325–32. [PubMed]