Expression profiling of adrenocortical neoplasms suggests a molecular signature of malignancy
David Velázquez-Fernández, MD, MSc,a Cecilia Laurell, MSc,d Janos Geli, MD,b Anders Höög, MD,“ Jacob Odeberg, MD, PhD,d Magnus Kjellman, MD, PhD,a Joakim Lundeberg, PhD,d Bertil Hamberger, MD, PhD,ª Peter Nilsson, PhD,d and Martin Bäckdahl, MD, PhD,a Stockholm, Sweden
Background. Distinguishing between adrenocortical adenomas and carcinomas is often difficult. Our aim was to investigate the differences in transcriptional profiles between benign and malignant adrenocortical neoplasms using complementary DNA microarray techniques.
Methods. We studied 7 patients with adrenocortical carcinomas and 13 with adenomas. Histopathology was reviewed in all patients; clinical follow-up was at least 1 year. Hybridizations were performed in duplicate against RNA reference. Expression levels were analyzed in the R environment for statistical computing with the use of aroma, limma, statistics, and class packages.
Results. Transcriptional profiles were homogeneous among adenomas, while carcinomas were much more heterogeneous. Hierarchical clustering and self-organizing maps could separate clearly carcinomas from adenomas. Among genes that were most significantly upregulated in carcinomas were 2 ubiquitin-related genes (USP4 and UFDIL) and several insulinlike growth factor-related genes (IGF2, IGF2R, IGFBP3 and IGFBP6). Among genes that were most significantly downregulated in carcinomas were a cytokine gene (CXCL10), several genes related to cell metabolism (RARRES2, ALDH1A1, CYBRD1 and GSTA4), and the cadherin 2 gene (CDH2).
Conclusions. Through the use of cDNA arrays, adrenocortical adenomas and carcinomas appear to be clearly distinguishable on the basis of their specific molecular signature. The biologic importance of the up- and downregulated genes is yet to be determined. (Surgery 2005;138:1087-94.)
From the Departments of Surgery” and Human Pathology,” and the Center for Molecular Medicine,b Karolinska University Hospital; and the Department of Biotechnology, Department of Gene Technology,“ KTH - Royal Institute of Technology
ADRENOCORTICAL NEOPLASMS are classified as adeno- mas or carcinomas. Adrenocortical carcinomas are relatively rare neoplasms, with a yearly incidence of 0.5 to 2 cases per million inhabitants. These neo- plasms are aggressive with a poor prognosis, with mean survival time of less than 12 months for unre- sectable neoplasms; the overall 5-year survival rate has been estimated as less than 40%. In contrast,
Presented at the 26th Annual Meeting of the American Associ- ation of Endocrine Surgeons, Cancun, Mexico, April 3-5, 2005. Supported by a grant from Cancerföreningen i Stockholm (Sweden), Swedish Research Council (Sweden), and the Con- sejo Nacional de Ciencia y Tecnología (CONACYT, México).
Drs Velázquez-Fernández and Laurell contributed equally to this article.
Reprint requests: Martin Backdahl, MD, Department of Surgery, Karolinska University Hospital, Stockholm, Sweden. E-mail: martin.backdahl@karolinska.se.
0039-6060/$ - see front matter
benign adrenocortical neoplasms are much more frequent and have a better prognosis.2
Distinguishing between adenomas and carcino- mas is often difficult. Adenomas usually appear as small, homogeneous, encapsulated masses, rarely exceeding 5 cm as their greatest diameter or 50 g in weight, respectively. Carcinomas usually are larger, weigh more, and frequently have necrosis, hemor- rhage, as well as vascular and/or capsular invasion.2
Tumor size has a positive correlation with malig- nancy and is regarded as the most important fea- ture to predict malignant behavior.2,3 A complete surgical resection is currently the only potentially curative intervention of the malignant neoplasms. Other forms of treatment such as radio or chemo- therapy have only a modest impact on patient survival.4
Using comparative genomic hybridization, we detected no genetic alterations in adenomas with a diameter less than 5 cm, whereas larger ade- nomas and carcinomas showed an increased num- ber of aberrations.3 The most common genetic
| ID | Histopathologic diagnosis | Age | Gender | Follow-up (mo) | Weight (g) | Size (cm) |
|---|---|---|---|---|---|---|
| 501 | Adenoma | 50 | F | 181 | 2 | |
| 885 | Adenoma | 38 | F | 151 | 12 | 2.5 |
| 1077 | Adenoma | 47 | M | 144 | 2.5 | |
| 1105 | Adenoma | 29 | F | 143 | 9 | 2 |
| 1180 | Adenoma | 63 | F | 140 | 36 | 5 |
| 1191 | Adenoma | 63 | M | 140 | 32 | 4 |
| 1196 | Adenoma | 79 | F | 112* | 7 | 1.5 |
| 1764 | Adenoma | 42 | F | 110 | 17 | 2.5 |
| 2302 | Adenoma | 64 | F | 86 | 26 | 4 |
| 2414 | Adenoma | 63 | F | 72 | 22 | 4 |
| 2915 | Adenoma | 66 | M | 61 | 22 | 4 |
| 3669 | Adenoma | 54 | F | 31 | 29 | 4 |
| 3793 | Adenoma | 59 | F | 28 | 19 | 2.5 |
| 50 | Carcinoma | 63 | F | 6+ | 1500 | 20 |
| 1559 | Carcinoma | 72 | M | 120 | 1036 | 11 |
| 1780 | Carcinoma | 56 | F | 9+ | 309 | 9 |
| 2582 | Carcinoma | 54 | F | 4+ | 1339 | 15 |
| 2716 | Carcinoma | 68 | M | 69 | 1420 | 15 |
| 3072 | Carcinoma | 68 | M | 53 | 660 | 12 |
| 3410 | Carcinoma | 84 | F | 32+ | 1700 | 19 |
*Dead from unrelated cause.
+Dead from disease.
aberrations found were gains on chromosomes 4 and 5, as well as losses on chromosomes 2, 11, and 17. These findings were confirmed by several independent groups.” 5-7
Over the past few years, genomic profiling tools, such as complementary DNA (cDNA) microarrays, have provided the opportunity to reveal transcrip- tional signatures, enabling prediction of malignant behavior in several different neoplasms.8,9 This technology also can provide a valuable insight into the molecular mechanisms involved in cancer development and progression, thereby helping to identify genes of potential therapeutic and diag- nostic importance. The purpose of this study was to determine if the expression profiling of adre- nocortical neoplasms could provide a distinction between adenomas and carcinomas.
MATERIAL AND METHODS
This study included 20 sporadic adrenocortical neoplasms, including 13 small adenomas (<5 cm) and 7 large carcinomas (>9 cm), selected from a cohort of patients operated on at the Karolinska University Hospital from 1986 to 2003. Median age of the group was 62 years (range, 29-84 years). There were 14 women and 6 men. Four of 7 patients with carcinomas died during follow-up with a me- dian survival time of 32 months. All patients with adenoma except 1 were still alive at the time the study was conducted (median survival, 121 months;
P <. 01). The death of this patient was not related to adrenal disease. Clinical and tumor characteristics are shown in Table I. The diagnosis of adrenocorti- cal carcinoma was based on histopathologic fea- tures: tumor size, capsular and vascular invasion, mitotic index, necrosis, and pleomorphism.
This study was approved by the ethical commit- tee at the Karolinska University Hospital. Informed consent was obtained from every patient.
RNA preparation. A fresh frozen piece of spec- imen weighing approximately 100 mg was used for RNA extraction. Representative sections from all specimens were subjected to histopathologic evalu- ation for neoplasm representativity. All samples were estimated to consist of more than 70% neo- plastic cells. Total RNA was extracted with the use of Qiagen mini- and midipreps kits for RNA purifica- tion. Universal human reference RNA (Stratagene, La Jolla, Calif) was used as a common RNA reference for all hybridizations. RNA quantity and quality were assessed by spectophotometry and Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA).
Microarray fabrication. The produced micro- arrays consisted of 29760 cDNA fragments spotted onto Ultra GAPS slides (Corning, Corning, NY) with a QArray (Genetix, New Milton, UK). Detailed protocols and the complete gene list can be found at http://www.biotech.kth.se/molbio/microarray.
Target labelling. The cDNA target was generated by reverse transcription of about 20 µg of total RNA
A
1100 ADENOMA 1191 ADENOMA
3793 ADENOMA
1764 ADENOMA
2302 ADENOMA
2414 ADENOMA
SOLS ADENOMA
3669 ADENOMA
BES ADENOMA
1196 ADENOMA
SOL ADENOMA
1077 ADENOMA
1105 ADENOMA
ISSS CANCER
1700 CANCER
SSOA CANCER
2716 CANCER
3072 CANCER SO CANCER
3410 CANCER
B
1100 ADENOMA 1191 ADENOMA
3793 ADENOMA
1764 ADENOMA
=303 ADENOMA
=414 ADENOMA
=915 ADENOMA
3669 ADENOMA
005 ADENOMA
1196 ADENOMA
SOL ADENOMA
1077 ADENOMA
1105 ADENOMA
1559 CANCER
1700 CANCER
SSO3 CANCER
2716 CANCER
3072 CANCER
SO CANCER
3410 CANCER
Insulin-like growth factor 2 (somatomedin A)
Nucleolar protein 1, 120kDa
Protein tyrosine phosphatase, receptor type
Mid-1-related chloride channel 1
Insulin-like growth factor 2 receptor Insulin-like growth factor 2 (somatomedin A)
SMT3 suppressor of mif two 3 homolog 1 (yeast)
Insulin-like growth factor 2 (somatomedin A’
“Myeloid/lymphoid or mixed-lineage leukemia
Zinc finger protein 32 (KOX 30)
Cadherin 13, H-cadherin (heart)
Baculoviral IAP repeat-containing 3
NGFI-A binding protein 1 (EGRI binding protein 1) Full-length cDNA clone CSODB007YNO1 Transmembrane protein
Ubiquitin specific protease 4 (proto-oncogene)
Ubiquitin fusion degradation 1-like
Aquaporin 3
Transcribed locus
Hypothetical protein PP1057
Inositol polyphosphate phosphatase-like 1
H3 histone, family 3B (H3.3B)
Insulin-like growth factor binding protein 6
Homeo box D9
Protocadherin alpha 6
Deleted in azoospermia 4
Mix interactor
Fibroblast growth factor receptor 4
Growth arrest-specific 2 like 3
Ring finger protein (C3HC4 type) 159
Thymopoietin
Cadherin 2, type 1, N-cadherin (neuronal)
Aldehyde dehydrogenase 1 family, member Al
ATP-binding cassette, sub-family G (WHITE), member 1
Cytochrome b reductase 1
Leucine rich repeat neuronal 3
Cytochrome b reductase 1
Spectrin, beta, non-erythrocytic 1
CDNA: FLJ21418 fis, clone COL04072
AE binding protein 1
Chemokine (C-X-C motif) ligand 10
Retinoic acid receptor responder (tazarotene induced)
Glutathione S-transferase A4
Potassium inwardlv-rectifving channel
18
1
8
Fold Change
in a 30 uL reaction with the use of 5 µg random hexamers (Operon), first-strand buffer (50 mmol/L TRIS HCl, pH 8.3; 75 mmol/L KCl; 3 mmol/L MgCl2), 0.01 mmol/L DTT, 400 units Superscript II (Invitrogen, Carlsbad, Calif), 2 mmol/L d(A/G/ C)TPs, 1.6 mmol/L dTTP (Amersham Biosciences, Uppsala, Sweden) and 0.4 mmol/L aminoallyl- dUTP (Sigma-Aldrich, St. Louis, Mo). After purifi- cation of the cDNA, monofunctional NHS-ester Cy3 or Cy5 fluorophores (Amersham Biosciences) were coupled to the amino-allyl groups. Neoplasm sam- ples were labeled with Cy5 and RNA reference with Cy3. Labeling quality and quantity were assessed with the use of a nanophotometer (Nanodrop Technologies, Wilmington, Del).
Hybridization, washing, and scanning. Slides were prehybridized in 5X saline sodium citrate (SSC), 0.1% sulfonyl dodecylsulfate (SDS), and 1% bovine serum albumin (Sigma-Aldrich) for 30 minutes at 42℃. The hybridization buffer used for all experiments contained 50% formamide, 5X SSC, 0.1% SDS, 10 µg COT-1 DNA (Invitrogen), and 5 µg transfer RNA (Sigma-Aldrich). The
hybridizations were carried out in hybridization chambers (Corning) at 42℃ for at least 18 hours. Posthybridization washes were carried out accord- ing to Corning’s recommendations. Hybridizations were performed in duplicate. The arrays were scanned at 532 and 635 nm with 10-um resolution with the use of a G2565BA DNA microarray scan- ner (Agilent Technologies).
Data analysis. Image analysis was performed in GenePix 5.1 software (Axon Instruments, Sun- nyvale, CA). After removal of bad quality spots (if less than 70% of foreground pixels were below background intensity plus 2 SDs in both channels or if the difference between ratio of medians and regression ratio exceeded 20% in one of the chan- nels), the remaining intensities were print tip LOW- ESS normalized in the R environment for statistical computing (http://www.R-project.org) with the aroma package (http://www.maths.lth.se/publica tions). B-test for ranking differentially expressed genes was performed on the average of technical replicates on genes represented in all samples with the use of the Limma package.1º Clustering was
| Name | Symbol | Genbank acc no. | UG cluster | Cytoband | Log2 ratio | Fold- change | t | ₱ | B | Log Odd |
|---|---|---|---|---|---|---|---|---|---|---|
| Upregulated | ||||||||||
| Ubiquitin-specific | USP4 | AA454143 | Hs.403828 | 3p21.3 | 5,90 | 59,69 | 21,03 | 1,41E-12 | 27,08 | 1.0 |
| protease 4 | ||||||||||
| (protooncogene) | ||||||||||
| Ubiquitin fusion | UFD1L | T57841 | Hs.404525 | 22q11.21 | 5,31 | 39,64 | 20,69 | 1,41E-12 | 26,77 | 1.0 |
| degradation 1-like | ||||||||||
| Inositol polyphosphate phosphatase-like 1 | INPPL1 | AA279072 | Hs.75339 | 11q23 | 4,50 | 22,65 | 19,88 | 2,25E-12 | 25,99 | 1.0 |
| Aquaporin 3 | AQP3 | R92737 | Hs.234642 | 9p13 | 4,97 | 31,30 | 18,60 | 7,05E-12 | 24,69 | 1.0 |
| H3 histone, family 3B (H3.3B) | H3F3B | AA608514 | Hs. 180877 | 17q25 | 4,09 | 17,05 | 18,26 | 8,39E-12 | 24,32 | 1.0 |
| Myeloid/lymphoid or mixed-lineage leukemia (trithorax homolog) | MLLT10 | N68327 | Hs.446451 | 10p12 | 5,51 | 45,62 | 17,49 | 1,74E-11 | 23,48 | 1.0 |
| Protein tyrosine | PPFIA1 | N49751 | Hs.503095 | 11q13.3 | 6,29 | 78,05 | 17,17 | 2,21E-11 | 23,11 | 1.0 |
| phosphatase, receptor type, f polypeptide (PTPRF), interacting protein (liprin), alpha 1 | ||||||||||
| Zinc finger protein 32 (KOX 30) | ZNF32 | T47230 | Hs.78765 | 10q22-q25 | 5,26 | 38,37 | 16,21 | 5,81E-11 | 21,97 | 1.0 |
| Cadherin 13, H-cadherin (heart) | CDH13 | R41787 | Hs.129340 | 16q24.2- | 4,19 | 18,24 | 15,44 | 1,45E-10 | 21,01 | 1.0 |
| q24.3 | ||||||||||
| Mid-1-related chloride channel 1 | MCLC | H51262 | Hs.534441 | 1p13.3 | 6,22 | 74,74 | 14,53 | 3,91E-10 | 19,81 | 1.0 |
| Insulinlike growth factor 2 receptor | IGF2R | T62547 | Hs.76473 | 6q26 | 6,30 | 78,92 | 14,37 | 4,38E-10 | 19,59 | 1.0 |
| Insulinlike growth factor | IGF2 | N74623 | Hs.251664 | 11p15.5 | 8,75 | 431,59 | 14,33 | 4,38E-10 | 19,53 | 1.0 |
| 2 (somatomedin A) | ||||||||||
| Insulinlike growth factor | IGF2 | N54596 | Hs.251664 | 11p15.5 | 7,08 | 135,41 | 14,31 | 4,38E-10 | 19,51 | 1.0 |
| 2 (somatomedin A) | ||||||||||
| Baculoviral IAP repeat-containing 3 | BIRC3 | H48533 | Hs.127799 | 11q22 | 3,49 | 11,27 | 14,17 | 5,05E-10 | 19,31 | 1.0 |
| Transcribed locus | AA939251 | Hs.528979 | 6,25 | 75,88 | 14,04 | 5,47E-10 | 19,13 | 1.0 | ||
| Nucleolar protein 1, 120kDa | NOL1 | N50854 | Hs.15243 | 12p13 | 8,35 | 325,23 | 13,67 | 8,46E-10 | 18,61 | 1.0 |
| Downregualted | ||||||||||
| Chemokine (C-X-C motif) ligand 10 | CXCL10 | AA878880 | Hs.413924 | 4q21 | -2,27 | 4,84 | 15,32 | 1,55E-10 | 20,86 | 1.0 |
| Retinoic acid receptor responder (tazarotene induced) 2 | RARRES2 | AA482067 | Hs.37682 | 7q36.1 | -2,59 | 6,01 | -14,10 | 5,23E-10 | 19,22 | 1.0 |
| Aldehyde dehydrogenase 1 family, member A1 | ALDH1A1 | AA664101 | Hs.76392 | 9q21.13 | -3,63 | 12,38 | 12,08 | 7,60E-09 | 16,20 | 1.0 |
| Cytochrome b reductase 1 | CYBRD1 | N54788 | Hs.31297 | 2q31.1 | -2,05 | 4,14 | 11,21 | 2,71E-08 | 14,76 | 1.0 |
| Glutathione | GSTA4 | AA152347 | Hs.169907 | 6p12.1 | -2,55 | 5,86 | -11,04 | 3,58E-08 | 14,46 | 1.0 |
| S-transferase A4 | ||||||||||
| Cadherin 2, type 1, | CDH2 | W49619 | Hs.334131 | 18q11.2 | -3,74 | 13,32 | -10,85 | 4,68E-08 | 14,14 | 1.0 |
| N-cadherin neuronal) |
| Name | Symbol | Genbank acc no. | UG cluster | Cytoband | Log2 ratio | Fold- change | t | ₱ | B | Log Odd |
|---|---|---|---|---|---|---|---|---|---|---|
| ATP-binding cassette, sub-family G (WHITE), member 1 | ABCG1 | R39446 | Hs.369055 | 21q22.3 | -2,67 | 6,36 | -10,70 | 6,00E-08 | 13,87 | 1.0 |
| Cytochrome b reductase 1 | CYBRD1 | AA150422 | Hs.31297 | 2q31.1 | -2,94 | 7,69 | -10,52 | 7,92E-08 | 13,55 | 1. 1.0 |
| Spectrin, beta, nonerythrocytic 1 | SPTBN1 | AA936302 | Hs.516006 | 2p21 | -1,97 | 3,92 | -10,43 | 9,06E-08 | 13,39 | 1.0 |
| Leucine rich repeat neuronal 3 | LRRN3 | AA452138 | Hs.3781 | 7q31.1 | -2,41 | 5,33 | -10,35 | 1,01E-07 | 13,24 | 1.0 |
| CDNA: FLJ21418 fis, clone COL04072 | AA417905 | Hs.205401 | -1,85 | 3,61 | -9,86 | 2,35E-07 | 12,36 | 1.0 | ||
| AE-binding protein 1 | AEBP1 | AA490462 | Hs.439463 | 7p13 | -1,58 | 3,00 | -9,55 | 3,93E-07 | 11,76 | 1.0 |
| Potassium | KCNJ8 | AA436184 | Hs.102308 | 12p11.23 | 2,92 | 7,58 | -9,52 | 4,11E-07 | 11,70 | 1.0 |
| inwardly-rectifying channel, subfamily J, member 8 | ||||||||||
| Metastasis associated family, member 3 | MTA3 | W80761 | Hs.435413 | 2p21 | -1,42 | 2,68 | -9,42 | 4,75E-07 | 11,52 | 1.0 |
| Calcium/calmodulin- dependent protein kinase I | CAMK1 | H29322 | Hs.434875 | 3p25.3 | -1,76 | 3,39 | -9,32 | 5,53E-07 | 11,31 | 1.0 |
GenBank acc no., GenBank accession number; UG cluster, UniGene Cluster designation number; Foldchange, fold expression change according to the M value (log 2 Ratio) t, moderated Student t test and its corresponding P value, adjusted by Benjamini and Hochberg method; B, Bayesian test value, Log Odd according to the B value that means the probability for a gene to be real differentially expressed.
performed with the statistics and class packages (http://www.R-project.org). Hierarchical cluster- ing of genes for visualisation of expression patterns was performed in MultiExperiment Viewer (MEV) (http://www.tigr.org/). Functional gene classification was performed according to Gene Ontology (http:// www.geneontology.org/GO.doc.html). Fisher exact test was used to evaluate significant overrepresenta- tion of GO terms and chromosomal locations among differentially expressed genes. To systemati- cally associate the gene expression profiles with the chromosome and cytogenetic gene locations, we de- cided to include the respective LocusLink identifi- cation number of the differentially regulated genes in Onto-Express (http://vortex.cs.wayne.edu). This software queries the National Center for Bio- technology (NCBI) map viewer and retrieves the number of genes on each chromosome.
RESULTS
Pattern of gene expression profiles. A total of 9670 spots/cDNA fragments were present in all samples in at least 1 technical replicate after quality filtering. In Figure 1A, the expression profiles of all samples were visualized in a heat map dendogram in which the genes have been hierarchically clus- tered with the use of Euclidean distances and aver- age linkage to provide an overview of the general pattern. The expression profiles among adenomas
are rather homogeneous and differ remarkably from the more heterogeneous profiles observed in carcinomas. When clustering on samples with the same dataset, adenomas and carcinomas were placed in two separate clusters, which revealed that the malignant phenotype indeed has an ex- pression profile distinguishable from the benign one. The same result was obtained with three differ- ent clustering techniques; hierarchical clustering, k-means clustering, and self-organizing maps.
Differentially expressed genes. An empirical Bayesian method (B-test) was applied to test for differential expression between adenomas and carcinomas. Of the 9760 genes, 571 were found to be differentially expressed (ie, fulfilling the criteria B > 0 in the B-test, more than a 2-fold differential expression and P < . 01 in a moderated t test). Among the 571 differentially expressed genes, 273 genes were upregulated, and 298 genes were downregulated in carcinomas, compared with adenomas. Significance scores and degrees of dif- ferential expression are listed in Table II for the 15 most significant upregulated genes and the 15 most downregulated genes. Expression profiles of the 50 most significant differentially expressed genes are displayed in Figure 1B across all neo- plasm samples to visualize the variation of this par- ticular subset of genes within and between neoplasm phenotype groups. Several insulinlike
6
DIGF2
@IGF2
= IGF2
4
= IGF2R
= IGFBP6
= IGFBP3
2
0
2
Y
9
00
AD 1180
AD 1191
AD 3793
AD 1764
AD 2302
AD 2414
AD 2915
AD 3669
AD 885
AD 1196
AD 501
AD 1077
AD 1105
CA 1559
CA 1780
CA 2582
CA 2716
CA 3072
CA 50
CA 3410
growth factors (IGFs) and 2 ubiquitin-related genes were among the 15 most significant upregu- lated genes. Figure 2 demonstrates the relative levels of IGF2, insulinlike growth factor 2 receptor (IGF2R), and insulinlike growth factor-binding proteins 3 and 6 (IGFBP3, IGFBP6) transcripts in all neoplasm samples. Their expression pattern clearly differs between adenomas and carcinomas.
Functional classification of the genes. A majority of upregulated genes in carcinomas appeared to have a predominant nuclear (GO 5634) location (P < . 004), while the downregulated genes had their highest representation in the cytoplasm (GO 5737) and plasma membrane (GO 5886) lo- cation (P < . 05). These findings were confirmed when biologic process ontology and molecular function ontology were used.
Chromosomal location. To investigate if certain chromosomes were overrepresented among the differentially expressed genes, we evaluated the chromosomal distribution of the 571 most signif- icantly affected genes. One hundred eighty-eight of the 273 upregulated and 218 of the 298 down- regulated genes belonged to unique LocusLink identification number, and were used for chromo- somal location. Significance of overrepresentation was assessed with Fisher exact test. Chromosomes 12 and 5 among upregulated genes, and chromo- somes 2 and 1 among downregulated genes in carcinomas were overrepresented.
DISCUSSION
Distinguishing between adrenocortical adeno- mas and carcinomas can be difficult. Previous studies have demonstrated that adrenocortical carcinomas display specific and more numerous
chromosomal and/or genetic alterations than ad- enomas.3,11 which could suggest a characteristic malignant genotype.
In the present study, we used a human cDNA microarray for determining a wider expression genomic profile. We selected 20 neoplasms; 13 small adenomas and 7 large carcinomas.
Expression profiles were shown to be clearly different between adenomas and carcinomas, with a homogeneous pattern among the adenomas and a heterogeneous pattern in the carcinomas. Of the 9670 genes, 571 exhibited a statistically significant difference in the expression level between adeno- mas and carcinomas. Functional classification of the upregulated genes in carcinomas showed pre- dominantly a nuclear location, while the down- regulated genes displayed cytoplasmic and plasma membrane location (P < . 05). Regarding chromo- somal localization of the most significantly upregu- lated genes in carcinomas, chromosomes 12 and 5 had a major number of involved loci, while chro- mosomes 2 and 1 had the major number of in- volved loci for the downregulated genes. These findings are consistent with earlier studies of com- parative genomic hybridization.3,5,7
The two most significant differentially upregulated genes were the ubiquitin-related genes: ubiquitin- specific protease 4 (USP4) and ubiquitin fusion degradation 1-like (UFD1L). Both genes were upregulated at least 40 times in carcinomas, com- pared with adenomas, and had the highest statistical significant Pand B values. USP4 is a protooncogene mapped to 3p21.3 and related to a group of deu- biquitinating enzymes located mainly in the cell nucleus. The overexpression of the USP4 gene is associated with malignant transformation of small cell adenocarcinoma cells in vitro and in vivo.12
The second ubiquitin-related gene is the UFD1L, which is mapped to 22q11.21. The product of this gene is involved in the degradation of ubiq- uitinated proteins; it is probably involved in chroma- tin remodeling.13 However, the precise biochemical role of UFD1L in human cells remains to be further elucidated. Inhibition of ubiquitin-related proteins has recently been suggested as a novel therapeutic strategy in human cancers. Different approaches have been proposed: for example, the use of specific inhibitors such as bortezomib (Velcade or PS-341) alone or in combination with chemotherapy (borte- zomib plus gemcitabine and carboplatin). A DNA vaccine, which includes an encoding tyrosinase- related protein 2 (neoplasm antigen) with an N-terminal-fused ubiquitin protein, also has been suggested to be useful. All these therapies are under evaluation.
An important group of overexpressed genes in this study includes members of the IGF system (IGF2, IGF2R, IGFBP3, and IGFBP6). The IGF- related transcripts levels across samples are showed in Figure 2. These genes are upregulated in carci- nomas, compared with adenomas. One of these, IGF2 is the most consistently found upregulated gene in adrenocortical carcinomas.14,15 The IGF2, mapped to 11p15.5, has an important role as an autocrine regulator of cell proliferation and apoptosis. The overexpression of IGF2 also has been associated with the progression of several other cancer types such as meningioma,16 and he- patocellular17 and colorectal cancer.18 Moreover, human antibodies directed to IGF2 protein inhibit prostate tumor growth in vivo. Furthermore, high plasma levels of IGF2 and low plasma levels of IGFBP3 are associated with a high risk of endome- trial cancer.19
Several reports have pointed out the IGF2 molecule as a potential target for therapy in human cancer. Advocated treatments include use of a growth hormone receptor antagonist (B2036- PEG), a ligand-specific antibody to human IGF- 1 and 2 (KM1468), and DNA-based therapy with the use of transcriptional regulatory sequences of IGF2 (plus a fragment of a diphtheria toxin gene for cytotoxicity). Some combinations of chemo- therapy agents (such as tamoxifen plus etoposide) also have been proposed. All these therapies have a considerable but variable effect in human neo- plasms expressing predominantly IGF2.
IGF2R is mapped to 6q25-q27 and is considered a tumor-suppressor gene. This gene encodes for a nonmitogenic receptor, which targets the IGF2 to the lysosomes for degradation; in this way, this gene controls IGF2 proliferative activity. Upregulation of the IGF2R gene in the carcinomas in the present study is somewhat contradictory to its role as a tumor-suppressor gene. However, studies20 have shown aberrant transcript levels associated with dif- ferent IGF2R mutations related to the development of hepatocellular and gastric carcinoma as well as adrenocortical neoplasms.
IGFBP6 is also is overexpressed in all the carci- nomas, a finding consistent with observations in testicular and papillary thyroid carcinoma.
IGFBP3 showed a lower level of upregulation among the carcinomas, compared with other IGF- related genes. IGFBP3 is mapped to 7p14-p12, and its product functions as a major carrier protein for circulating IGF2, thereby blocking its activity. A negative correlation has been reported be- tween cancer risk and IGFBP3.19,21 These data have been challenged by other groups.22 IGFBP3
overexpression may be induced by the cyclooxy- genase 2 inhibitor Celecoxib, leading to apoptosis in vitro.23
One example of downregulated genes in carci- nomas, compared with adenomas, is the chemo- kine (C-X-C motif) ligand 10 (CXCL10), which is involved mainly in cell communication and mor- phogenesis. CXCL10 is mapped to 4q21 and encodes an angiogenesis and inflammatory-modu- lating cytokine. In contrast with our results, CXCL10 has been found overexpressed in colo- rectal cancer.24
Other downregulated genes such as the retinoic acid receptor responder 2 (RARRES2; tazarotene- induced); the aldehyde dehydrogenase 1 family, member A1 (ALDH1A1); the cytochrome b reduc- tase 1 (CYBRD1); and the glutathione S-transferase A4 (GSTA4) are involved mainly in cell metabo- lism. Cadherin 2, type 1, N-cadherin (CDH2; neuronal) is mapped to 18q11.2 and encodes a cell-adhesion molecule. This gene has been re- ported to be epigenetically silenced in pancreatic cancer,25 thereby decreasing its expression. This could potentially be the underlying mechanism for the down regulation observed in the neoplasm samples.
Through the use of cDNA arrays, adenomas and carcinomas may be distinguished clearly on the basis of their specific molecular signature. The biologic importance of the up- and downregulated genes has to be determined further. Identification of these differentially expressed genes may enhance our understanding of the molecular biology of the development of adrenocortical neoplasms and aid in creating new diagnostic and prognostic tools.
We thank Lisa Ånfalk, Fredrik Lyngman, Annelie Waldén, Daniel Johansson, Amilcar Flores, Ulla Enberg, and Catharina Larsson for technical assistance and valuable suggestions.
REFERENCES
1. Vassilopoulou-Sellin R, Schultz PN. Adrenocortical carci- noma. Clinical outcome at the end of the 20th century. Cancer 2001;92:1113-21.
2. Boushey RP, Dackiw AP. Adrenal cortical carcinoma. Curr Treat Options Oncol 2001;2:355-64.
3. Kjellman M, Kallioniemi OP, Karhu R, et al. Genetic aberra- tions in adrenocortical tumors detected using comparative genomic hybridization correlate with tumor size and malig- nancy. Cancer Res 1996;15;56:4219-23.
4. Dackiw AP, Lee JE, Gagel RF, et al. Adrenal cortical carci- noma. World J Surg 2001;25:914-26.
5. Dohna M, Reincke M, Mincheva A, et al. Adrenocortical carcinoma is characterized by a high frequency of chromo- somal gains and high-level amplifications. Genes Chromo- somes Cancer 2000;28:145-52.
6. Zhao J, Speel EJ, Muletta-Feurer S, et al. Analysis of ge- nomic alterations in sporadic adrenocortical lesions. Gain of chromosome 17 is an early event in adrenocortical tu- morigenesis. Am J Pathol 1999;155:1039-45.
7. Sidhu S, Marsh DJ, Theodosopoulos G, et al. Comparative genomic hybridization analysis of adrenocortical tumors. J Clin Endocrinol Metab 2002;87:3467-74.
8. Santin AD, Zhan F, Bellone S, et al. Gene expression pro- files in primary ovarian serous papillary tumors and normal ovarian epithelium: identification of candidate molecular markers for ovarian cancer diagnosis and therapy. Int J Cancer 2004;112:14-25.
9. Chevillard S, Ugolin N, Vielh P, et al. Gene expression pro- filing of differentiated thyroid neoplasms: diagnostic and clinical implications. Clin Cancer Res 2004;10:6586-97.
10. Smyth GK. Linear models and empirical Bayes methods for assessing differential expression in microarray experiments [serial online]. Stat Aplications Genet Mol Biol 2004;3(1).
11. Kjellman M, Roshani L, Teh BT, et al. Genotyping of adre- nocortical tumors: very frequent deletions of the MEN1 locus in 11q13 and of a 1-centimorgan region in 2p16. J Clin Endocrinol Metab 1999;84:730-5.
12. Gray DA, Inazawa J, Gupta K, et al. Elevated expression of Unph, a proto-oncogene at 3p21.3, in human lung tumors. Oncogene 1995;10:2179-83.
13. Amati F, Condo I, Conti E, et al. Analysis of intracellular dis- tribution and apoptosis involvement of the Ufd1l gene pro- duct by over-expression studies. Cell Biochem Funct 2003; 21:263-7.
14. Giordano TJ, Thomas DG, Kuick R, et al. Distinct transcrip- tional profiles of adrenocortical tumors uncovered by DNA microarray analysis. Am J Pathol 2003;162:521-31.
15. de Fraipont F, El Atifi M, Cherradi N, et al. Gene expression profiling of human adrenocortical tumors using cDNA mi- croarrays identifies several candidate genes as markers of ma- lignancy. J Clin Endocrinol Metab 2005 Mar;90(3):1819-29.
16. Wrobel G, Roerig P, Kokocinski F, et al. Microarray-based gene expression profiling of benign, atypical and anaplastic meningiomas identifies novel genes associated with menin- gioma progression. Int J Cancer 2005;114:249-56.
17. Li H, Zhang N. Study on the expression and genomic im- printing status of insulin-like growth factor 2 gene in hepa- tocellular carcinoma. Zhonghua Gan Zang Bing Za Zhi 2004;12:347-9.
18. Nosho K, Yamamoto H, Takamaru H, et al. A case of colo- rectal carcinoma in adenoma analyzed by a cDNA array. Int J Colorectal Dis 2005 Aug 2; [Epub ahead of print].
19. Oh JC, Wu W, Tortolero-Luna G, et al. Increased plasma levels of insulin-like growth factor 2 and insulin-like growth factor binding protein 3 are associated with endometrial can- cer risk. Cancer Epidemiol Biomarkers Prev 2004;13:748-52.
20. Leboulleux S, Gaston V, Boulle N, et al. Loss of heterozygos- ity at the mannose 6-phosphate/insulin-like growth factor 2 receptor locus: a frequent but late event in adrenocortical tumorigenesis. Eur J Endocrinol 2001;144:163-8.
21. Deal C, Ma J, Wilkin F, et al. Novel promoter polymorphism in insulin-like growth factor-binding protein-3: correlation with serum levels and interaction with known regulators. J Clin Endocr Metab 2001;86:1274-80.
22. Renehan AG, Zwahlen M, Minder C, et al. Insulin-like growth factor (IGF)-I, IGF binding protein-3, and cancer risk: systematic review and meta-regression analysis. Lancet 2004;363:1346-53.
23. Levitt RJ, Buckley J, Blouin MJ, et al. Growth inhibition of breast epithelial cells by celecoxib is associated with upregu- lation of insulin-like growth factor binding protein-3 expres- sion. Biochem Biophys Res Commun 2004;316:421-8.
24. Zhang R, Zhang H, Zhu W, et al. Mob-1, a Ras target gene, is overexpressed in colorectal cancer. Oncogene 1997;14:1607-10.
25. Hagihara A, Miyamoto K, Furuta J, et al. Identification of 27 5’ CpG islands aberrantly methylated and 13 genes silenced in human pancreatic cancers. Oncogene 2004;23:8705-10.