ttp://www.bsava.com
PETSAVERS PAPER
Investigation of prognostic indicators for human uveal melanoma as biomarkers of canine uveal melanoma metastasis P. Malho*, K. Dunn†, D. Donaldson*, R. R. Dubielzig‡, Z. Birand§ and M. Starkey§ *Comparative Ophthalmology Unit, Animal Health Trust, Kentford, Newmarket, CB8 7UU, UK †FOCUS-EyePathLab, Murarrie, Queensland 4172, Australia ‡Comparative Ocular Pathology Laboratory, University of Wisconsin-Madison, Madison, WI 53706, USA §Molecular Oncology Group, Animal Health Trust, Kentford, Newmarket CB8 7UU, UK
OBJECTIVE: To evaluate if 14 genes that discriminate metastasising and non-metastasising human uveal melanomas can differentiate metastasising and non-metastasising uveal melanomas in dogs. METHODS: Nineteen archival biopsies of eyes with a histopathological classification of primary benign (n=9) and malignant (n=10) uveal melanoma were selected. Thoracic and/or abdominal metastases confirmed metastatic spread of the primary tumour in seven dogs during the follow-up period. Gene expression was assayed by Reverse Transcription-quantitative Polymerase Chain Reaction. Genes displaying statistically significant differences in expression between the metastasising and non-metastasising tumours were identified. RESULTS: Four genes (HTR2B, FXR1, LTA4H and CDH1) demonstrated increased expression in the metastasising uveal melanomas. CLINICAL SIGNIFICANCE: This preliminary study illustrates the potential utility of gene expression markers for predicting canine uveal melanoma metastasis. The genes displaying elevated expression in the metastasising tumours are part of a 12-discriminating gene set used in a routine assay, performed on fine needle aspirate biopsies collected without enucleation, for predicting human uveal melanoma metastasis. Further work is required to validate the results. Journal of Small Animal Practice (2013) 54, 584–593 DOI: 10.1111/jsap.12141 Accepted: 9 September 2013
INTRODUCTION Uveal melanoma is the most common intraocular tumour found in dogs and humans (Bussanich et al. 1987, Egan et al. 1988). These tumours originate from malignant transformation of melanocytes in the uvea. The anterior uvea (iris and ciliary body) is the most frequently recognised site for the development of uveal melanoma in dogs (Trucksa 1983), in contrast Presented in abstract form at the 43rd Annual Meeting of the American College of Veterinary Ophthalmologists, Portland, October 2012.
584
to humans where the choroid is affected in more than 90% of cases (Damato 2010). The first abnormality commonly observed with uveal melanoma is a dark-pigmented mass in the anterior segment (Bussanich et al. 1987). Large ciliary body tumours are often seen through the pupil, as they displace the iris and cause dyscoria. Amelanotic uveal melanomas are rare (Dubielzig 1990). The potential for development of metastatic disease is a concern for both human and canine patients. In the dog, the malignant form accounts for 20% of all uveal melanomas and haematogenous spread results in liver and lung metastases in 4 to 8% of all cases (Wilcock & Peiffer 1986, Bussanich et al. 1987, Giuliano et al. 1999). At first presentation, metastases are usually
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
Biomarkers of canine uveal melanoma metastasis
not present, or at least not clinically detectable (micrometastases). A study of four cases of canine uveal melanoma (Wilcock & Peiffer 1986) reported that metastases became evident during the 3 months post-presentation, and that for all cases death resulted from metastatic disease, or humane euthanasia, 6 months postdiagnosis. The metastatic subpopulation of malignant uveal melanomas represents a prognostic challenge as there are no clinical or histological features that accurately predict tumour spread. Although a significant difference in survival time between histologically benign and malignant cases has been demonstrated (Giuliano et al. 1999), neither high mitotic index (the defining histological feature of malignant uveal melanomas) nor tumour size and extension, are able to differentiate metastasising from non-metastasising malignant uveal melanomas (Giuliano et al. 1999). In the absence of a prognostic test that can be performed on eyes in situ, eyes with unrecognised non-metastasising uveal melanomas may undergo premature enucleation in an unnecessary attempt to prevent metastatic spread. Over 50% of human patients with choroidal melanomas die because of metastatic disease in the 15 years following diagnosis (Gamel et al. 1993). Once there is evidence of metastatic disease death usually occurs within 7 months (Gamel et al. 1993). Although poor survival has been correlated with advanced patient age, anterior tumour location, increased tumour size, epithelioid cell type and tumoural scleral invasion, they are inaccurate at predicting metastasis in an individual patient (Prescher et al. 1996, Scholes et al. 2003, Worley et al. 2007, Gill et al. 2012). Cytogenetic analysis has identified a number of unbalanced genomic aberrations associated with patient outcome (Prescher et al. 1996, Sisley et al. 1997, Kilic et al. 2005, van Gils et al. 2008). Subsequently, transcriptional profiling was shown to stratify human uveal melanomas into two groups – “Class 1” is associated with a
good prognosis and “Class 2” with a high risk of metastatic death (Onken et al. 2004, 2006, Petrausch et al. 2008, van Gils et al. 2008). Furthermore, a gene expression signature-based upon 12 genes is sufficient for accurate classification (Onken et al. 2010, 2012). Measurement of the expression of these genes in tumour biopsies collected by fine needle aspiration (Onken et al. 2006, 2012) is frequently used in prognostication for human choroidal melanoma and avoids enucleation. The aim of this study was to investigate potential biomarkers of canine uveal melanoma metastasis. It was hypothesised that one or more of 14 genes which discriminate human uveal melanomas on the basis of metastatic potential will display differences in expression between metastasising and non-metastasising canine uveal melanomas. Ultimately, the results of this study demonstrate the potential for developing a gene expression signature-based clinical test to accurately predict uveal melanoma metastasis in the dog.
MATERIALS AND METHODS To adhere to the MIQE Guidelines for the publication of quantitative Polymerase Chain Reaction (qPCR) experiments (Bustin et al. 2009), additional details are provided in the Supporting Information. Tissue sample selection Nineteen formalin-fixed paraffin-embedded (FFPE) archival canine primary uveal melanomas with a histopathological classification (see below) of benign (n=9) and malignant (n=10) were selected (Table 1). In addition to histopathological diagnosis, the inclusion criterion was the absence of concurrent
Table 1. Uveal melanoma samples ID
Breed
Gender
Age (Years)
Eye
Ocular involvement
Mitotic index (Number of mitoses/10 highpower fields)
Histopathological diagnosis
Metastasis
UM1 UM2 UM3 UM4 UM5 UM6 UM7 UM8 UM9 UM10 UM11
Collie cross Labrador retriever Flat-coated retriever Weimaraner Bernese Mountain dog Flat-coated retriever Labrador retriever Miniature schnauzer Labrador retriever Border collie Jack Russell terrier
ME ME MN FE FE FS FS FE MN FS FS
2 14 6 027 5 11·66 13·08 12 11·42 11·33 10
OS OD OD OD OS OS OS OD OS OD OD
Iris Iris, ciliary body Iris, ciliary body Iris Iris Iris, ciliary body Iris, ciliary body Choroid Choroid Iris, ciliary body Iris
0 0-1 0-1 0-1 0-1 0-1 0-1 0-1 0-1 4-7 5-100
Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Malignant melanoma Malignant melanoma
UM12 UM13 UM14 UM15 UM16 UM17 UM18 UM19
Crossbreed Labrador retriever Shih-tzu Labrador retriever Shih-tzu cross Golden retriever Labrador retriever cross Fox terrier
MN FE MN FS FS MN FS MN
17·58 10·5 6 7·92 1 9·5 9 9
OD OD OS OS OD OD OS OS
Iris, ciliary body Iris, ciliary body Iris, ciliary body Choroid Iris, ciliary body Iris, ciliary body Iris, ciliary body Iris, ciliary body
10-100 100-200 40-80 27 48 24-100 46 100-150
Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma
None None None None None None None None None Systemic Liver and pulmonary None Pulmonary None None Pulmonary Pulmonary Pulmonary Pulmonary
Follow-up (Months)
27 8 16 16 17 22 9 11* 19 8* 5* 6 14* 3* 21 9* 3* 8* 5*
FE Entire female, FS Spayed female, ME Entire male, MN Neutered male, OD oculus dextrum, OS oculus sinistrum The follow-up interval post-ocular tissue sampling is detailed *Indicates the length of the follow-up period at the time of death
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
585
P. Malho et al.
neoplasia at the time of ocular excisional biopsy or enucleation, as verified by complete physical examination, thoracic radiographs and abdominal ultrasound. A minimum clinical followup after ocular tissue sampling of 6 months was required unless the dog died, or was humanely euthanased. Metastatic disease was confirmed by radiography (cases UM11, UM13, UM16, UM17, UM18 and UM19) and/or necropsy (cases UM10, UM11 and UM19). In the absence of evidence of concurrent neoplasms, metastatic spread of the primary uveal melanoma was concluded. Histological evaluation of the tumour samples FFPE ocular tissues were stained with haematoxylin and eosin and reviewed by (KD) one of the authors. Bleached slides were prepared for heavily pigmented tumours and Melan A immunohistochemistry performed for definitive diagnosis of tumours without detectable pigment. The diagnosis of malignant melanoma was based on the identification of ≥4 mitotic figures per 10 high-power fields in one or more areas of a tumour, along with evidence of cellular and nuclear pleomorphism and atypia (Wilcock & Peiffer 1986). Selection of candidate discriminating (“target”) genes Fourteen genes were selected based on their consistent difference in expression between metastasising and non-metastasising human uveal melanomas (Table 2). Nine of the genes are also located on sections of canine chromosomes syntenic to human chromosomal regions affected by metastatic-death-associated
aneuploidies in uveal melanomas (Table 2). Eleven of the 14 genes are part of a 12-discriminating gene set (Onken et al. 2010) used in a clinical assay for predicting metastasis in human uveal melanomas. There is currently no recognised canine orthologue for the twelfth gene (ID2) in the set. Design of quantitative Polymerase Chain Reaction (qPCR) assays Hydrolysis probe (TaqMan; Life Technologies) quantitative PCR (qPCR) assays (Table 3) were designed (Beacon Designer; PREMIER Biosoft) to amplify 70 to 120 bp sequences that (where possible) span exon–exon boundaries. For genes predicted to encode alternate transcripts, assay design was based upon a common region, whilst transcript regions that share significant sequence similarity (as assessed by BLAST search) with nonhomologous mRNAs were excluded as targets. Isolation of total RNA and reverse transcription Total RNA was isolated (RecoverAll Total Nucleic Acid Isolation Kit; Life Technologies) from each FFPE melanoma and quantified by fluorometry (Quant-iT RiboGreen RNA Reagent; Life Technologies). Reverse transcription was performed using 120 ng of total RNA and the High-Capacity cDNA Reverse Transcription Kit (Life Technologies). For assay of gene expression, random primers were employed for reverse transcription, and two reactions, one with (“+RT”), and one without (“−RT”), reverse transcriptase were performed for each melanoma. For assay of RNA integrity, reverse transcription was initiated using a glucose-6-phosphate isomerase-specific primer.
Table 2. Genes assayed Gene symbol
Human chromosome
Cytogenetic band
Syntenic dog chromosome(s)
Chromosomal location of canine orthologue
Canine Ensembl gene ID*
Inclusion criteria
Up-regulated class†
NFIA ECM1 HTR2B ROBO1 EIF1B SATB1 LMCD1 FXR1 NEDD9 MTUS1 ENPP2 LTA4H CDH1 RAB31
1 1 2 3 3 3 3 3 6 8 8 12 16 18
p31.3 q21·3 q37·1 p12·2 p22·1 p24·3 p26·1 q26·33 p24·2 p22 q24·12 q23·1 q22·1 p11·22
5, 6, 17 38, 17, 7 37, 25 20, 31 20, 23 20, 23 20, 23 23, 34 35, 12 25, 16 13 26, 10, 15 5, 2 1, 7
5 17 25 31 23 23 20 34 35 16 13 15 5 7
ENSCAFG00000018848 ENSCAFG00000012022 ENSCAFG00000010964 ENSCAFG00000024476 ENSCAFG00000005119 ENSCAFG00000005839 ENSCAFG00000005570 ENSCAFG00000011582 ENSCAFG00000009773 ENSCAFG00000006979 ENSCAFG00000000865 ENSCAFG00000006440 ENSCAFG00000020305 ENSCAFG00000018727
A,‡ C§ A,|| B** A,††,‡ B** A,†† B,** C‡,§,‡‡,§§ A,|| B,** C‡,§,‡‡,§§ A,|| B,** C‡,§,‡‡,§§ A,††,¶¶, ‡, B,** C‡,§,‡‡,§§ A,|| B,** C§,‡‡,§§ A,‡ C|||| A,||,¶¶,‡ B,** C*** A,||,††,‡ C§§ A,|| B** A,|| B** A,††,‡ B**
?¶ C2 C2 C1 C1 C1 C1 C1 C1 C1 C1 C1 C2 C2
A Differential expression between metastasising and non-metastasising human uveal melanomas, B Member of a 12 discriminating gene set intended for use in a routine clinical assay for predicting metastasis in human uveal melanomas, C Located on canine chromosomal segments syntenic to human chromosome regions whose aneuploidy is associated with uveal melanoma metastasis and poor survival *Gene ID. as annotated in the CanFam 3.0 genome assembly (Ensembl Dog Genome Browser) † Molecular class in which expression is increased: C1 Human uveal melanoma class associated with a low risk of metastasis, C2 Human uveal melanoma class associated with a high risk of metastasis ¶ NFIA (cDNA FLJ39164 fis, clone OCBBF2002656) is a member of a classifier gene list for human uveal melanomas that have a high risk of metastasis although it is not clear from the publication whether it has an elevated or reduced level of expression in these tumours |||| Prescher et al. (1995) ‡‡ Prescher et al. (1996) §§ Sisley et al. (1997) || Onken et al. (2004) § Kilic et al. (2005) †† Onken et al. (2006) ***Onken et al. (2008) ¶¶ Petrausch et al. (2008) ‡ van Gils et al. (2008) **Onken et al. (2010)
586
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
Biomarkers of canine uveal melanoma metastasis
Table 3. Quantitative PCR assays TaqMan Gene Ensembl Transcript ID* (No. Duplex symbol of alternate transcripts†) 1 2 3 4 5 6 7
NFIA ECM1 HTR2B ROBO1 EIF1B SATB1 LMCD1 FXR1 NEDD9 MTUS1 ENPP2 LTA4H CDH1 RAB31
Probe (5-3)‡
Forward primer (5-3)
Reverse primer (5′-3′)
ENSCAFT00000029932 (4) 1-CATCGCCCATTATCCAGCAGC-3 CCGTCGACTCTTCATTTTC TGGACAAACTCTTTCAGC ENSCAFT00000019091 (1) 2-TCACAGCCACACAAACCGC-4 GAGACTAGATATTCCCGCTG CACAGAATCGGGTCATTG ENSCAFT00000038772 (2) 1-CATTGCCATTCCAGTCCCTATTAGA-3 CAGTGGTATGGTTAATTTC ACTCCGATCAGTTTCTA ENSCAFT00000012652 (1) 2-CCACTCTCTCCCCACCTTTGTA-4 GCAACTCTGAACTGTAA GGTCATCTTTGTCTGTC ENSCAFT00000008231 (1) 1-TCCTCTTTGACAATGCCAACCTC-3 CAAGGTGACCAAAGAAA CCATGAACCTTAAGCTG ENSCAFT00000009427 (1) 2-AGCAGGAATCTCCCAGGCA-4 GGTACGTGATGAACTGA TTCGGAGGATTTCTGAAA ENSCAFT00000008964 (2) 1-TGAGCTCCATGTACTGCAGACC-3 ACACCATCACCTATGAGTG CTGGCTGCTTCTCTTTTG ENSCAFT00000018461 (3) 2-ACATACAGCAAGCTAGGAAGGTTCC-4 TGGCAATTGGAACACATG TCAGCACTCTCTCCATAG ENSCAFT00000015519 (1) 1-TCTCCTTTTCCAACAGCTCCTTC-3 GGATGGATGATTATGACTAC GCTTGTTCTGTTTGATGA ENSCAFT00000039471 (2) 2-TTGGAAATAACATATGCCTCCAGCA-4 TCCCTGTTAAAATGGAAAG CCTTCTTTTCAGGGCTAA ENSCAFT00000001352 (2) 1-AGCCCAGGAGATCGCACA-3 CCAACGTTTAAGTACAAGAC GGGTTCCATTATTCGGAG ENSCAFT00000010410 (1) 2-ATCTCACAGCTGGACAGGAAATCT-4 GTCACTCTCCAATGTTATTG CCAAGTTTTGTTGGTCAC ENSCAFT00000032337 (3) 1-TGGATTAACCTCCAGCCAACCG-3 GGAACAGAGGATAACGTATC GAGTGAAAATGGCACCA ENSCAFT00000029725 (1) 2-TGGACCGTGCTCCTTCAGTT-4 TGCAGCTGTCATAGTATATG CAGCAATAGCCATTACAATG
Amplicon size (bp) 110 97 71 91 80 98 90 115 102 81 106 73 83 113
*Transcript ID. as annotated in the CanFam 3.0 genome assembly (Ensembl Dog Genome Browser) † TaqMan assay design used the listed transcript as template, but where a gene is predicted to encode alternate transcripts the target amplicon is located in a region common to all transcript variants ‡ Probe dyes: 1=Cy5, 2=ROX, 3=BHQ-3, 4=BHQ-2
RNA quality assessment Each RNA sample was assayed for DNA contamination via screening an aliquot of “−RT” reaction product by end-point PCR amplification of a 110bp exon-specific amplicon derived from one of the target genes (NFIA, ENSCAFG000000029932). A quantitative assessment of the integrity of each RNA sample (Penland et al. 2007) was obtained by hydrolysis probe qPCR measurement of the ratio of the copy number of a 3′-end sequence to that of a “more 5′-end” sequence at the 3′-end of the ubiquitous transcript glucose-6-phosphate isomerase (GPI; ENSCAFT00000011733). Following reverse transcription, the 5′-end and 3′-end GPI sequences were preamplified (TaqMan PreAmp Master Mix Kit; Life Technologies), and triplicate “duplex” qPCRs performed. Quantification cycles (Cq) were derived using the PCR machine software (Quansoft, Techne). The 3′-end and 5′-end GPI sequence copy numbers in each preamplified cDNA were derived from amplicon-specific standard curves, plotted from results obtained for serial dilutions of a preamplified cDNA prepared from “Dog Universal RNA” (BioChain). qPCR analysis of target genes To facilitate the reliable measurement of gene expression in low integrity FFPE RNAs, the target genes in 1 µL of each cDNA (“+RT” reaction product) were simultaneously preamplified for 14 cycles using the TaqMan PreAmp Master Mix Kit (Life Technologies; Li et al. 2008). Preamplified cDNAs were diluted 20-fold in 10mM Tris–HCl (pH 8·0), 1mM EDTA, and 5 µL used as the template in triplicate qPCR reactions. QPCR was performed using the QuantiTect Multiplex PCR NoROX Kit (Qiagen) and the Quantica Real-Time Nucleic Acid Detection System (Techne), thermocycling at 95°C for 15 minutes, followed by 50 cycles of 94°C for 1 minute, 57°C for 30 seconds, and 60°C for 1 minute. Quantification cycles (Cq) were derived using the PCR machine software (Quansoft, Techne), and a median Cq calculated for each gene-cDNA triplicate reaction series. Results were excluded if the Cq standard deviation (sd) Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
was >0·5. Genes with a median Cq of ≥35 were deemed not to be expressed. qPCR analysis of reference genes In order to facilitate multiple reference gene normalisation of the target gene expression data, the 19 melanoma cDNAs were also assayed for the expression of 10 genes (“Dog Housekeeping Genes” set; Qiagen) known to be constitutively expressed across a broad range of tissues and biological conditions. Prior to qPCR, the 10 reference cDNAs were preamplified (1 × RT2 PreAmp PCR Mastermix, Dog Housekeeping Genes RT2 PreAmp Pathway Primer Mix; Qiagen) for 8 cycles. Preamplified cDNAs were diluted with 111·75 µL of water and 1 µL aliquots used as templates in qPCR (Dog Housekeeping Gene RT2 qPCR Primer Assay, 1 × RT2 SYBR Green ROX qPCR Mastermix; Qiagen). Thermocycling at 95°C for 10 minutes followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute was performed using the StepOnePlus Real-Time PCR System (Life Technologies). The identity of each amplification product as a single species was confirmed by dissociation curve analysis. Relative quantification of target gene expression The median Cq measure of the expression of each target gene was converted to a relative measure of gene expression, the “Normalised Relative Quantity” (NRQ; Hellemans et al. 2007). Calculation of NRQs was performed using qbasePLUS2 (biogazelle), and was achieved using the reference genes found to demonstrate the lowest variability in expression across the eligible cDNAs (target gene Cq
PETSAVERS PAPER
Investigation of prognostic indicators for human uveal melanoma as biomarkers of canine uveal melanoma metastasis P. Malho*, K. Dunn†, D. Donaldson*, R. R. Dubielzig‡, Z. Birand§ and M. Starkey§ *Comparative Ophthalmology Unit, Animal Health Trust, Kentford, Newmarket, CB8 7UU, UK †FOCUS-EyePathLab, Murarrie, Queensland 4172, Australia ‡Comparative Ocular Pathology Laboratory, University of Wisconsin-Madison, Madison, WI 53706, USA §Molecular Oncology Group, Animal Health Trust, Kentford, Newmarket CB8 7UU, UK
OBJECTIVE: To evaluate if 14 genes that discriminate metastasising and non-metastasising human uveal melanomas can differentiate metastasising and non-metastasising uveal melanomas in dogs. METHODS: Nineteen archival biopsies of eyes with a histopathological classification of primary benign (n=9) and malignant (n=10) uveal melanoma were selected. Thoracic and/or abdominal metastases confirmed metastatic spread of the primary tumour in seven dogs during the follow-up period. Gene expression was assayed by Reverse Transcription-quantitative Polymerase Chain Reaction. Genes displaying statistically significant differences in expression between the metastasising and non-metastasising tumours were identified. RESULTS: Four genes (HTR2B, FXR1, LTA4H and CDH1) demonstrated increased expression in the metastasising uveal melanomas. CLINICAL SIGNIFICANCE: This preliminary study illustrates the potential utility of gene expression markers for predicting canine uveal melanoma metastasis. The genes displaying elevated expression in the metastasising tumours are part of a 12-discriminating gene set used in a routine assay, performed on fine needle aspirate biopsies collected without enucleation, for predicting human uveal melanoma metastasis. Further work is required to validate the results. Journal of Small Animal Practice (2013) 54, 584–593 DOI: 10.1111/jsap.12141 Accepted: 9 September 2013
INTRODUCTION Uveal melanoma is the most common intraocular tumour found in dogs and humans (Bussanich et al. 1987, Egan et al. 1988). These tumours originate from malignant transformation of melanocytes in the uvea. The anterior uvea (iris and ciliary body) is the most frequently recognised site for the development of uveal melanoma in dogs (Trucksa 1983), in contrast Presented in abstract form at the 43rd Annual Meeting of the American College of Veterinary Ophthalmologists, Portland, October 2012.
584
to humans where the choroid is affected in more than 90% of cases (Damato 2010). The first abnormality commonly observed with uveal melanoma is a dark-pigmented mass in the anterior segment (Bussanich et al. 1987). Large ciliary body tumours are often seen through the pupil, as they displace the iris and cause dyscoria. Amelanotic uveal melanomas are rare (Dubielzig 1990). The potential for development of metastatic disease is a concern for both human and canine patients. In the dog, the malignant form accounts for 20% of all uveal melanomas and haematogenous spread results in liver and lung metastases in 4 to 8% of all cases (Wilcock & Peiffer 1986, Bussanich et al. 1987, Giuliano et al. 1999). At first presentation, metastases are usually
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
Biomarkers of canine uveal melanoma metastasis
not present, or at least not clinically detectable (micrometastases). A study of four cases of canine uveal melanoma (Wilcock & Peiffer 1986) reported that metastases became evident during the 3 months post-presentation, and that for all cases death resulted from metastatic disease, or humane euthanasia, 6 months postdiagnosis. The metastatic subpopulation of malignant uveal melanomas represents a prognostic challenge as there are no clinical or histological features that accurately predict tumour spread. Although a significant difference in survival time between histologically benign and malignant cases has been demonstrated (Giuliano et al. 1999), neither high mitotic index (the defining histological feature of malignant uveal melanomas) nor tumour size and extension, are able to differentiate metastasising from non-metastasising malignant uveal melanomas (Giuliano et al. 1999). In the absence of a prognostic test that can be performed on eyes in situ, eyes with unrecognised non-metastasising uveal melanomas may undergo premature enucleation in an unnecessary attempt to prevent metastatic spread. Over 50% of human patients with choroidal melanomas die because of metastatic disease in the 15 years following diagnosis (Gamel et al. 1993). Once there is evidence of metastatic disease death usually occurs within 7 months (Gamel et al. 1993). Although poor survival has been correlated with advanced patient age, anterior tumour location, increased tumour size, epithelioid cell type and tumoural scleral invasion, they are inaccurate at predicting metastasis in an individual patient (Prescher et al. 1996, Scholes et al. 2003, Worley et al. 2007, Gill et al. 2012). Cytogenetic analysis has identified a number of unbalanced genomic aberrations associated with patient outcome (Prescher et al. 1996, Sisley et al. 1997, Kilic et al. 2005, van Gils et al. 2008). Subsequently, transcriptional profiling was shown to stratify human uveal melanomas into two groups – “Class 1” is associated with a
good prognosis and “Class 2” with a high risk of metastatic death (Onken et al. 2004, 2006, Petrausch et al. 2008, van Gils et al. 2008). Furthermore, a gene expression signature-based upon 12 genes is sufficient for accurate classification (Onken et al. 2010, 2012). Measurement of the expression of these genes in tumour biopsies collected by fine needle aspiration (Onken et al. 2006, 2012) is frequently used in prognostication for human choroidal melanoma and avoids enucleation. The aim of this study was to investigate potential biomarkers of canine uveal melanoma metastasis. It was hypothesised that one or more of 14 genes which discriminate human uveal melanomas on the basis of metastatic potential will display differences in expression between metastasising and non-metastasising canine uveal melanomas. Ultimately, the results of this study demonstrate the potential for developing a gene expression signature-based clinical test to accurately predict uveal melanoma metastasis in the dog.
MATERIALS AND METHODS To adhere to the MIQE Guidelines for the publication of quantitative Polymerase Chain Reaction (qPCR) experiments (Bustin et al. 2009), additional details are provided in the Supporting Information. Tissue sample selection Nineteen formalin-fixed paraffin-embedded (FFPE) archival canine primary uveal melanomas with a histopathological classification (see below) of benign (n=9) and malignant (n=10) were selected (Table 1). In addition to histopathological diagnosis, the inclusion criterion was the absence of concurrent
Table 1. Uveal melanoma samples ID
Breed
Gender
Age (Years)
Eye
Ocular involvement
Mitotic index (Number of mitoses/10 highpower fields)
Histopathological diagnosis
Metastasis
UM1 UM2 UM3 UM4 UM5 UM6 UM7 UM8 UM9 UM10 UM11
Collie cross Labrador retriever Flat-coated retriever Weimaraner Bernese Mountain dog Flat-coated retriever Labrador retriever Miniature schnauzer Labrador retriever Border collie Jack Russell terrier
ME ME MN FE FE FS FS FE MN FS FS
2 14 6 027 5 11·66 13·08 12 11·42 11·33 10
OS OD OD OD OS OS OS OD OS OD OD
Iris Iris, ciliary body Iris, ciliary body Iris Iris Iris, ciliary body Iris, ciliary body Choroid Choroid Iris, ciliary body Iris
0 0-1 0-1 0-1 0-1 0-1 0-1 0-1 0-1 4-7 5-100
Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Benign melanoma Malignant melanoma Malignant melanoma
UM12 UM13 UM14 UM15 UM16 UM17 UM18 UM19
Crossbreed Labrador retriever Shih-tzu Labrador retriever Shih-tzu cross Golden retriever Labrador retriever cross Fox terrier
MN FE MN FS FS MN FS MN
17·58 10·5 6 7·92 1 9·5 9 9
OD OD OS OS OD OD OS OS
Iris, ciliary body Iris, ciliary body Iris, ciliary body Choroid Iris, ciliary body Iris, ciliary body Iris, ciliary body Iris, ciliary body
10-100 100-200 40-80 27 48 24-100 46 100-150
Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma Malignant melanoma
None None None None None None None None None Systemic Liver and pulmonary None Pulmonary None None Pulmonary Pulmonary Pulmonary Pulmonary
Follow-up (Months)
27 8 16 16 17 22 9 11* 19 8* 5* 6 14* 3* 21 9* 3* 8* 5*
FE Entire female, FS Spayed female, ME Entire male, MN Neutered male, OD oculus dextrum, OS oculus sinistrum The follow-up interval post-ocular tissue sampling is detailed *Indicates the length of the follow-up period at the time of death
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
585
P. Malho et al.
neoplasia at the time of ocular excisional biopsy or enucleation, as verified by complete physical examination, thoracic radiographs and abdominal ultrasound. A minimum clinical followup after ocular tissue sampling of 6 months was required unless the dog died, or was humanely euthanased. Metastatic disease was confirmed by radiography (cases UM11, UM13, UM16, UM17, UM18 and UM19) and/or necropsy (cases UM10, UM11 and UM19). In the absence of evidence of concurrent neoplasms, metastatic spread of the primary uveal melanoma was concluded. Histological evaluation of the tumour samples FFPE ocular tissues were stained with haematoxylin and eosin and reviewed by (KD) one of the authors. Bleached slides were prepared for heavily pigmented tumours and Melan A immunohistochemistry performed for definitive diagnosis of tumours without detectable pigment. The diagnosis of malignant melanoma was based on the identification of ≥4 mitotic figures per 10 high-power fields in one or more areas of a tumour, along with evidence of cellular and nuclear pleomorphism and atypia (Wilcock & Peiffer 1986). Selection of candidate discriminating (“target”) genes Fourteen genes were selected based on their consistent difference in expression between metastasising and non-metastasising human uveal melanomas (Table 2). Nine of the genes are also located on sections of canine chromosomes syntenic to human chromosomal regions affected by metastatic-death-associated
aneuploidies in uveal melanomas (Table 2). Eleven of the 14 genes are part of a 12-discriminating gene set (Onken et al. 2010) used in a clinical assay for predicting metastasis in human uveal melanomas. There is currently no recognised canine orthologue for the twelfth gene (ID2) in the set. Design of quantitative Polymerase Chain Reaction (qPCR) assays Hydrolysis probe (TaqMan; Life Technologies) quantitative PCR (qPCR) assays (Table 3) were designed (Beacon Designer; PREMIER Biosoft) to amplify 70 to 120 bp sequences that (where possible) span exon–exon boundaries. For genes predicted to encode alternate transcripts, assay design was based upon a common region, whilst transcript regions that share significant sequence similarity (as assessed by BLAST search) with nonhomologous mRNAs were excluded as targets. Isolation of total RNA and reverse transcription Total RNA was isolated (RecoverAll Total Nucleic Acid Isolation Kit; Life Technologies) from each FFPE melanoma and quantified by fluorometry (Quant-iT RiboGreen RNA Reagent; Life Technologies). Reverse transcription was performed using 120 ng of total RNA and the High-Capacity cDNA Reverse Transcription Kit (Life Technologies). For assay of gene expression, random primers were employed for reverse transcription, and two reactions, one with (“+RT”), and one without (“−RT”), reverse transcriptase were performed for each melanoma. For assay of RNA integrity, reverse transcription was initiated using a glucose-6-phosphate isomerase-specific primer.
Table 2. Genes assayed Gene symbol
Human chromosome
Cytogenetic band
Syntenic dog chromosome(s)
Chromosomal location of canine orthologue
Canine Ensembl gene ID*
Inclusion criteria
Up-regulated class†
NFIA ECM1 HTR2B ROBO1 EIF1B SATB1 LMCD1 FXR1 NEDD9 MTUS1 ENPP2 LTA4H CDH1 RAB31
1 1 2 3 3 3 3 3 6 8 8 12 16 18
p31.3 q21·3 q37·1 p12·2 p22·1 p24·3 p26·1 q26·33 p24·2 p22 q24·12 q23·1 q22·1 p11·22
5, 6, 17 38, 17, 7 37, 25 20, 31 20, 23 20, 23 20, 23 23, 34 35, 12 25, 16 13 26, 10, 15 5, 2 1, 7
5 17 25 31 23 23 20 34 35 16 13 15 5 7
ENSCAFG00000018848 ENSCAFG00000012022 ENSCAFG00000010964 ENSCAFG00000024476 ENSCAFG00000005119 ENSCAFG00000005839 ENSCAFG00000005570 ENSCAFG00000011582 ENSCAFG00000009773 ENSCAFG00000006979 ENSCAFG00000000865 ENSCAFG00000006440 ENSCAFG00000020305 ENSCAFG00000018727
A,‡ C§ A,|| B** A,††,‡ B** A,†† B,** C‡,§,‡‡,§§ A,|| B,** C‡,§,‡‡,§§ A,|| B,** C‡,§,‡‡,§§ A,††,¶¶, ‡, B,** C‡,§,‡‡,§§ A,|| B,** C§,‡‡,§§ A,‡ C|||| A,||,¶¶,‡ B,** C*** A,||,††,‡ C§§ A,|| B** A,|| B** A,††,‡ B**
?¶ C2 C2 C1 C1 C1 C1 C1 C1 C1 C1 C1 C2 C2
A Differential expression between metastasising and non-metastasising human uveal melanomas, B Member of a 12 discriminating gene set intended for use in a routine clinical assay for predicting metastasis in human uveal melanomas, C Located on canine chromosomal segments syntenic to human chromosome regions whose aneuploidy is associated with uveal melanoma metastasis and poor survival *Gene ID. as annotated in the CanFam 3.0 genome assembly (Ensembl Dog Genome Browser) † Molecular class in which expression is increased: C1 Human uveal melanoma class associated with a low risk of metastasis, C2 Human uveal melanoma class associated with a high risk of metastasis ¶ NFIA (cDNA FLJ39164 fis, clone OCBBF2002656) is a member of a classifier gene list for human uveal melanomas that have a high risk of metastasis although it is not clear from the publication whether it has an elevated or reduced level of expression in these tumours |||| Prescher et al. (1995) ‡‡ Prescher et al. (1996) §§ Sisley et al. (1997) || Onken et al. (2004) § Kilic et al. (2005) †† Onken et al. (2006) ***Onken et al. (2008) ¶¶ Petrausch et al. (2008) ‡ van Gils et al. (2008) **Onken et al. (2010)
586
Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
© 2013 British Small Animal Veterinary Association
Biomarkers of canine uveal melanoma metastasis
Table 3. Quantitative PCR assays TaqMan Gene Ensembl Transcript ID* (No. Duplex symbol of alternate transcripts†) 1 2 3 4 5 6 7
NFIA ECM1 HTR2B ROBO1 EIF1B SATB1 LMCD1 FXR1 NEDD9 MTUS1 ENPP2 LTA4H CDH1 RAB31
Probe (5-3)‡
Forward primer (5-3)
Reverse primer (5′-3′)
ENSCAFT00000029932 (4) 1-CATCGCCCATTATCCAGCAGC-3 CCGTCGACTCTTCATTTTC TGGACAAACTCTTTCAGC ENSCAFT00000019091 (1) 2-TCACAGCCACACAAACCGC-4 GAGACTAGATATTCCCGCTG CACAGAATCGGGTCATTG ENSCAFT00000038772 (2) 1-CATTGCCATTCCAGTCCCTATTAGA-3 CAGTGGTATGGTTAATTTC ACTCCGATCAGTTTCTA ENSCAFT00000012652 (1) 2-CCACTCTCTCCCCACCTTTGTA-4 GCAACTCTGAACTGTAA GGTCATCTTTGTCTGTC ENSCAFT00000008231 (1) 1-TCCTCTTTGACAATGCCAACCTC-3 CAAGGTGACCAAAGAAA CCATGAACCTTAAGCTG ENSCAFT00000009427 (1) 2-AGCAGGAATCTCCCAGGCA-4 GGTACGTGATGAACTGA TTCGGAGGATTTCTGAAA ENSCAFT00000008964 (2) 1-TGAGCTCCATGTACTGCAGACC-3 ACACCATCACCTATGAGTG CTGGCTGCTTCTCTTTTG ENSCAFT00000018461 (3) 2-ACATACAGCAAGCTAGGAAGGTTCC-4 TGGCAATTGGAACACATG TCAGCACTCTCTCCATAG ENSCAFT00000015519 (1) 1-TCTCCTTTTCCAACAGCTCCTTC-3 GGATGGATGATTATGACTAC GCTTGTTCTGTTTGATGA ENSCAFT00000039471 (2) 2-TTGGAAATAACATATGCCTCCAGCA-4 TCCCTGTTAAAATGGAAAG CCTTCTTTTCAGGGCTAA ENSCAFT00000001352 (2) 1-AGCCCAGGAGATCGCACA-3 CCAACGTTTAAGTACAAGAC GGGTTCCATTATTCGGAG ENSCAFT00000010410 (1) 2-ATCTCACAGCTGGACAGGAAATCT-4 GTCACTCTCCAATGTTATTG CCAAGTTTTGTTGGTCAC ENSCAFT00000032337 (3) 1-TGGATTAACCTCCAGCCAACCG-3 GGAACAGAGGATAACGTATC GAGTGAAAATGGCACCA ENSCAFT00000029725 (1) 2-TGGACCGTGCTCCTTCAGTT-4 TGCAGCTGTCATAGTATATG CAGCAATAGCCATTACAATG
Amplicon size (bp) 110 97 71 91 80 98 90 115 102 81 106 73 83 113
*Transcript ID. as annotated in the CanFam 3.0 genome assembly (Ensembl Dog Genome Browser) † TaqMan assay design used the listed transcript as template, but where a gene is predicted to encode alternate transcripts the target amplicon is located in a region common to all transcript variants ‡ Probe dyes: 1=Cy5, 2=ROX, 3=BHQ-3, 4=BHQ-2
RNA quality assessment Each RNA sample was assayed for DNA contamination via screening an aliquot of “−RT” reaction product by end-point PCR amplification of a 110bp exon-specific amplicon derived from one of the target genes (NFIA, ENSCAFG000000029932). A quantitative assessment of the integrity of each RNA sample (Penland et al. 2007) was obtained by hydrolysis probe qPCR measurement of the ratio of the copy number of a 3′-end sequence to that of a “more 5′-end” sequence at the 3′-end of the ubiquitous transcript glucose-6-phosphate isomerase (GPI; ENSCAFT00000011733). Following reverse transcription, the 5′-end and 3′-end GPI sequences were preamplified (TaqMan PreAmp Master Mix Kit; Life Technologies), and triplicate “duplex” qPCRs performed. Quantification cycles (Cq) were derived using the PCR machine software (Quansoft, Techne). The 3′-end and 5′-end GPI sequence copy numbers in each preamplified cDNA were derived from amplicon-specific standard curves, plotted from results obtained for serial dilutions of a preamplified cDNA prepared from “Dog Universal RNA” (BioChain). qPCR analysis of target genes To facilitate the reliable measurement of gene expression in low integrity FFPE RNAs, the target genes in 1 µL of each cDNA (“+RT” reaction product) were simultaneously preamplified for 14 cycles using the TaqMan PreAmp Master Mix Kit (Life Technologies; Li et al. 2008). Preamplified cDNAs were diluted 20-fold in 10mM Tris–HCl (pH 8·0), 1mM EDTA, and 5 µL used as the template in triplicate qPCR reactions. QPCR was performed using the QuantiTect Multiplex PCR NoROX Kit (Qiagen) and the Quantica Real-Time Nucleic Acid Detection System (Techne), thermocycling at 95°C for 15 minutes, followed by 50 cycles of 94°C for 1 minute, 57°C for 30 seconds, and 60°C for 1 minute. Quantification cycles (Cq) were derived using the PCR machine software (Quansoft, Techne), and a median Cq calculated for each gene-cDNA triplicate reaction series. Results were excluded if the Cq standard deviation (sd) Journal of Small Animal Practice
•
Vol 54
•
November 2013
•
was >0·5. Genes with a median Cq of ≥35 were deemed not to be expressed. qPCR analysis of reference genes In order to facilitate multiple reference gene normalisation of the target gene expression data, the 19 melanoma cDNAs were also assayed for the expression of 10 genes (“Dog Housekeeping Genes” set; Qiagen) known to be constitutively expressed across a broad range of tissues and biological conditions. Prior to qPCR, the 10 reference cDNAs were preamplified (1 × RT2 PreAmp PCR Mastermix, Dog Housekeeping Genes RT2 PreAmp Pathway Primer Mix; Qiagen) for 8 cycles. Preamplified cDNAs were diluted with 111·75 µL of water and 1 µL aliquots used as templates in qPCR (Dog Housekeeping Gene RT2 qPCR Primer Assay, 1 × RT2 SYBR Green ROX qPCR Mastermix; Qiagen). Thermocycling at 95°C for 10 minutes followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute was performed using the StepOnePlus Real-Time PCR System (Life Technologies). The identity of each amplification product as a single species was confirmed by dissociation curve analysis. Relative quantification of target gene expression The median Cq measure of the expression of each target gene was converted to a relative measure of gene expression, the “Normalised Relative Quantity” (NRQ; Hellemans et al. 2007). Calculation of NRQs was performed using qbasePLUS2 (biogazelle), and was achieved using the reference genes found to demonstrate the lowest variability in expression across the eligible cDNAs (target gene Cq
