- Email: [email protected]
Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation
Pathology (- 2017) -(-), pp. 1–7
A N ATO M I C A L PAT H O L O G Y
Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation NIMA MESBAH ARDAKANI1,2, CARLA THOMAS1, CLEO ROBINSON1,2, KYM MINA2,3, NATHAN TOBIAS HARVEY1,2, BENHUR AMANUEL1,2 AND BENJAMIN ANDREW WOOD1,2 1
Department of Anatomical Pathology, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, 2University of Western Australia, School of Pathology and Laboratory Medicine, Crawley, and 3Department of Diagnostic Genomics, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, WA, Australia
Summary Distinction between melanocytic naevi and melanoma occasionally poses a diagnostic challenge in ambiguous cases showing overlapping histological features. Melanomas are characterised by the presence of multiple genomic copy number variants (CNVs), while this is not a feature of naevi. We assessed the feasibility and utility of array-based comparative genomic hybridisation (aCGH) to assess CNVs in melanocytic lesions. DNA was extracted from formalin fixed, paraffin embedded (FFPE) sections of unambiguous naevi (n = 19) and melanomas (n = 19). The test DNA and gender mismatched human reference DNA were differentially labelled with fluorophores. Equal quantities of the two DNA samples were mixed and cohybridised to a SurePrint G3 Human CGH 8x60K array, and digitally scanned to capture and quantify the relative fluorescence intensities. The ratio of the fluorescence intensities was analysed by Cytogenomics software (Agilent). Frequent large CNVs were identified in 94.7% of melanoma samples, including losses of 9p (73.6%), 9q (52.6%), 10q (36.8%), 11q (36.8%), 3p (21%), and 10p (21%), and gains of 6p (42.1%), 7p (42.1%), 1q (36.8%), 8q (31.5%) and 20q (21%). Only one naevus showed two large copy number changes. Overall aCGH showed a specificity and sensitivity of 94.7% in separating naevi from melanomas. Based on our results, aCGH can be successfully used to analyse CNVs of melanocytic lesions utilising FFPE derived biopsy samples, providing a potentially useful adjunctive test for the classification of diagnostically challenging melanocytic proliferations. Key words: Comparative genomic hybridisation; copy number variations; melanoma; melanocytic naevi; virtual karyotyping. Received 26 June, revised 24 October, accepted 7 November 2016 Available online: xxx
INTRODUCTION Distinction of melanocytic naevi from cutaneous melanoma by histopathological examination is a common component of the everyday workload for many pathologists. While
morphological criteria for this separation are relatively well established, in practice these are not always easily applicable, and there are occasional ambiguous lesions characterised by overlapping features. This is reflected in several studies demonstrating only moderate interobserver agreement for the diagnosis of melanocytic lesions1,2 while others have reported that up to 11% of melanocytic diagnoses are significantly changed after specialist review.3 Misdiagnosis can lead to significant under- or overtreatment, and a false negative diagnosis of cutaneous melanoma remains the most common reason for malpractice claims against pathologists.4 Thus, there exists a clear need for ancillary testing to aid in the distinction between benign and malignant melanocytic proliferations in the setting of ambiguous histological features. The potential for cytogenetic aberrations to provide a diagnostic tool for separating melanoma from benign proliferations has been the subject of previous investigations.5,6 Gains and losses of genetic material are common in melanomas, but (with the exception of specific single abnormalities in Spitz naevi) not in melanocytic naevi.7 Indeed melanoma is notorious for showing the most unstable genome amongst all human solid tumours.8 Customised fluorescence in situ hybridisation (FISH) probes have been developed based on common chromosomal changes described in melanomas; however, these probes only assess limited loci and cannot interrogate the whole genome.9 Array-based comparative genomic hybridisation (aCGH) can detect chromosomal gains and losses of variable size across the entire genome.10 Despite the attractiveness of aCGH as a diagnostic tool, its use in clinical practice is extremely limited, in part due to significant technical requirements in regard to assay optimisation and validation for formalin fixed, paraffin embedded (FFPE) tissue and the need for microdissection to enable isolation of relatively pure lesional DNA.11 We assessed the feasibility and utility of aCGH in detection of CNVs in a cohort of unambiguous melanocytic lesions.
MATERIALS AND METHODS Selection of cases This study was approved as part of an intradepartmental routine quality and development exercise. A total of 45 samples of FFPE tissue reported as
Print ISSN 0031-3025/Online ISSN 1465-3931 Crown Copyright © 2017 Published by Elsevier B.V. on behalf of Royal College of Pathologists of Australasia. All rights reserved. DOI: http://dx.doi.org/10.1016/j.pathol.2016.11.008 Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
2
Pathology (2017), -(-), -
MESBAH ARDAKANI et al.
unambiguous malignant melanoma (MM) or melanocytic naevus were identified from the archives of Department of Anatomical Pathology, PathWest Laboratory Medicine (QEII Medical Centre, Perth, Western Australia). All cases were reviewed by two dermatopathologists (NMA and BAW) to confirm the original classification based on routine histological criteria. Two haematoxylin and eosin (H&E) stained slides representative of tumour blocks, one before and the other after 15 unstained slides, were examined and marked to select areas of viable tumour with maximum lesional cell content. These were then micro-dissected by scraping off 5 mm sections on glass slides using a scalpel blade. To be included in this study, lesions were required to have a minimum tumour thickness of 1 mm (using the Breslow methodology). Normal male and female genomic DNA (Promega, Australia) was used as reference DNA for array CGH experiments. Genomic DNA extraction Genomic DNA was extracted from FFPE tissue using the Qiagen DNeasy Blood and Tissue Kit, with modifications recommended by the ULS labelling system manufacturer (Agilent, Australia). Briefly, approximately 15 microdissected sections were heat deparaffinised at 90 C, followed by overnight treatment with 1 M sodium thiocyanate. This was followed by 48 h proteinase K treatment and then RNase A treatment. DNA was then purified using the Qiagen DNeasy Blood and Tissue Kit (Agilent), according to the manufacturer’s instructions but substituting the wash buffer AW2 with 80% ethanol and eluting the DNA in nuclease-free water. Extracted DNA was quantified by spectrophotometry using a NanoDrop ND-2000 (NanoDrop, USA). Ratio of absorbance at 260/280 was used to assess DNA purity, and samples with a ratio of ~1.80 were regarded as sufficiently pure and suitable for ULS labelling. All DNA samples were visualised on 1.0% agarose gel and fragment sizes were assessed against a 1 kb DNA ladder. DNA labelling DNA was labelled using an optimised version of the protocol for ULS labelling of FFPE DNA (Agilent). Prior to labelling, reference and FFPE DNA was heat fragmented at 95 C for 10 and 2 min, respectively, then 250 ng of tumour and reference DNA was chemically labelled by incubating with 0.5 mL of ULS-Cy3 and Cy5, respectively, in a thermal cycler with a heated lid for 30 min. Un-reacted dye was removed using KREApure filters (Agilent). The degree of labelling (DoL) was determined according to the manufacturer’s recommendations using the NanoDrop ND-2000. DoL values between 0.75% and 2.5% were regarded as optimal for Cy5, while values between 1.75% and 3.5% were optimal for Cy3-labelled DNA. Array hybridisation and scanning Cy3-labelled tumour DNA was combined with an equivalent amount of Cy5labelled sex mismatched reference DNA. Repetitive sequences were blocked with human Cot-1 DNA (Invitrogen, USA) and samples were hybridised onto SurePrint G3 Human CGH Microarrays, 8x60K (Agilent) according to manufacturer’s instructions. Following hybridisation for 40 h, microarray slides were washed according to manufacturer’s instructions and stored under nitrogen before being scanned on a DNA Microarray Scanner (Agilent). Data analysis Scanned images were analysed using Feature Extraction software (Agilent), which normalises the fluorescent intensity of both dyes at each probe and calculates their ratio, expressed on a logarithmic scale (probe log2 ratio). It also computes a set of quality control (QC) metrics, including the average green and red signal intensity at all the probes as well as the background signal (noise) and signal-to-noise ratio using non-hybridising control probes. Feature extracted data were then analysed using CytoGenomics Software (Agilent). FISH analysis Five mm thick, unstained, FFPE tissue sections were subject to in situ hybridisation with melanoma four-colour FISH kit including probes for RREB1 (6p25), CEP 6 (6p11.1-q11.1), MYB (6q23), and CCND1 (11q13) as well as CDKN2A/CEP 9 FISH probe kit (Vysis, USA).11,12 A standard hybridisation protocol was followed, as previously reported.9 A minimum of 60 tumour nuclei were counted for FISH analysis.
SNP array analysis Ten unstained sections from FFPE material were obtained for microdissection, with a single prior and subsequent H&E stained section used to confirm adequacy of tumour tissue and to mark areas for microdissection. DNA was extracted using a Qiagen extraction kit according to the manufacturer’s instruction (Qiagen, Germany). The Infinium HD assay was performed utilising Illumina iScan and HumanCytoSNP FFPE-12 BeadChip array according to the manufacturer’s protocol (Illumina, USA). The BeadChips were stained, and then imaged, using a BeadArray Reader (Illumina). Image data were analysed with GenomeStudio (Illumina).
RESULTS Bio-analytical adjustment aCGH primary data were analysed according to the Aberration Detection Method 2 (ADM-2) algorithm as previously described.10 The sensitivity threshold for the ADM-2 algorithm was adjusted based on self-self experiments performed on two samples. The post-analytic filters were also rectified based on two intra-array experiments. The optimal sensitivity threshold for the ADM-2 algorithm was defined as 6.1 after testing multiple thresholds. In addition, a minimum of 10 consecutive probes was established as the minimum postanalytic filter for CNVs which resulted in identical calls for intra-array duplicates performed on two samples. The log2 ratio threshold for detecting gains and losses of genomic material was set at 0.25. Quality control (QC) Opposite sex reference DNA was hybridised against each test DNA as an internal control for the quality and validity of results of each experiment. If the expected differences of the sex chromosomes (chromosome X and Y) were not met in a given sample, the experiment was considered unsatisfactory. QC metrics were also assessed for each sample. If QC metrics, in particular derivative log ratio (DLR) spread and red or green signal intensity, were not within the accepted limits, the experiment was considered unsatisfactory. A DLR spread of less than 0.5 was considered optimal, however DLR spread of equal or less than 0.6 was also accepted if the sex chromosome patterns were as expected. The samples with unsatisfactory results were repeated if there was adequate residual DNA or optimal remaining tissue to re-extract DNA. Post-analytic interpretation Copy number variants (CNVs) detected in each sample were first cross referenced with the Database of Genomic Variants (DGV) to exclude potential polymorphic CNVs.13 Large CNVs involving at least 100 consecutive probes were defined as pathogenic if CNVs overlapped exonic regions. CNVs involving less than 100 consecutive probes were studied individually. Variants with high log2 ratio detected by Cytogenomic software as homozygous deletions or amplifications were also considered significant if overlapped with areas on the genome harbouring genes known to be involved in melanomagenesis such as CCND1, CDKN2A, PTEN, P53, MYC, CDK4, NF1 and MDM2, or known to be frequently altered in melanomas such as MYB1 and RREB1.14 The remaining small (<100 probe) CNVs were considered as being of unknown significance.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
Analytical sensitivity of aCGH To evaluate the clonal detection threshold of the array, we serially diluted and tested a commercially available DNA sample with known germline heterozygous deletion of 13q13.3q21.33 (DNA sample NA01484; Coriell Institute, USA). Given the minimum average log2 ratio per probe was set at 0.25, the aCGH was expected to detect CNVs present in at least 33% of the constituent cells in a sample. Thus, we diluted the reference DNA by adding portions of normal DNA to achieve samples with 100%, 50%, 40%, 33%, 25% and 20% of abnormal DNA. The aCGH was able to detect the known deletion in samples with 100%, 50%, 40% and 33%, but not in samples with 25% and 20% of abnormal cells. Therefore the analytical sensitivity of the assay was established at expected 33% cut-off. Final study cohort After assessment of QC metrics and sex chromosome changes in each run, 12 samples showed technically unsatisfactory results. In five cases the quantity of remaining DNA was low and an attempt to re-extract DNA was unsuccessful due to low tumor cell content. Seven samples were repeated, of which five achieved satisfactory QC metrics on the repeat. The remaining cases were excluded from the study. After excluding the technically unsatisfactory results, 19 melanomas and 19 naevi were included in this study. The final cohort was composed of 38 lesions from 38 patients, including 24 men and 14 women with a median age of 58 years (range 15–86, mean 54.5 years). The patients in the melanoma cohort were older than the patients with benign naevi (mean age of 65.5 and 38.2 years, respectively, p < 0.05). The index lesions were thicker in the melanoma cohort with an average thickness of 6.65 mm as opposed to a mean thickness of 2.74 mm in naevi (p < 0.05). The melanoma subtypes included nine nodular, four superficial spreading, two acral lentiginous melanoma, one lentigo maligna melanoma and three metastatic melanomas. In metastatic melanomas the primary tumour was a superficial spreading melanoma in two cases, and in one case the primary tumour was unknown. In primary melanomas the Breslow thickness ranged from 1.1 mm to 15 mm with an average of 4.33 mm and a median of 3.5 mm. The melanomas were mainly located on the trunk and extremities while the naevi where predominantly from the head and neck areas. Histologically the naevi showed features of common compound or intradermal naevi which could be further classified as ‘Miescher’ naevus (n = 9), ‘Unna’ naevus (n = 3) and naevi with attributes of congenital naevus (n = 7) based on the pattern of growth. Detailed demographics are presented in Table 1. Genomic variants All but one case of melanoma demonstrated multiple large unbalanced CNVs. All but one of the melanocytic naevi showed either no CNVs or occasional small variants of unknown significance. One naevus showed two relatively large genomic losses on the long arms of chromosome 10 and 19, which were considered significant CNVs. One melanoma showed a normal aCGH profile. This case was a superficial spreading melanoma arising on the scalp of an elderly man with large areas of regression and a prominent lymphocytic
3
Table 1 Demographics of the patients and characteristics of the lesions in the study cohort Melanoma n = 19 Gender, n Male Female Age, years Range Average Thickness/depth of lesion, mm Range Average Median Melanoma subtype Superficial spreading Nodular Lentigo maligna Acral Other Site Head and neck Trunk Extremities
Naevi n = 19
12 7
12 7
23–86 65.5
15–78 38.2
1.55–18 6.65 5
1–5 2.74 2.75
4 9 1 2 3 3 9 7
10 8 1
infiltrate (Fig. 1). The sensitivity and specificity of aCGH in separation of unambiguous benign naevi from melanomas was 94.7% [confidence interval (CI) 71.8–99.7%]. The melanomas showed an average of 11.3 CNVs (range 0–25) with a mean number of 5.3 gains (range 0–22) and 6 losses (range 0–17). In most cases the losses were more numerous than gains; however, in four cases gains were predominant. One of these latter cases was an acral subungal melanoma with 22 gains and 3 losses. The most common CNVs in melanomas encompassed large losses of chromosome 9p (73.6%), 9q (52.6%), 10q (36.8%), 11q (36.8%), 3p (21%), and 10p (21%), and gains of chromosome 6p (42.1%), 7p (42.1%), 1q (36.8%), 8q (31.5%) and 20q (21%) (Fig. 2). These aberrations showed significant overlap with regions of the genome containing oncogenes or tumour suppressor genes known to be involved in melanomagenesis. For example, loss of chromosome 9p21 harbouring the CDKN2A (p16) gene was seen in 13 (68.4%) melanomas, four of which appeared to be homozygous deletions resulting in complete loss of tumour suppressor function. Other genes commonly involved in melanoma development such as RREB1, MYB, MYC and CDK4 were also well represented in our cohort. Further validation of CNVs The array CGH results were further validated by a four colour FISH assay (Vysis) with addition of a 9p21 (containing CDKN2A, p16 gene) probe (Vysis) in two melanoma cases. All aCGH calls within the detection range of the FISH assay were confirmed in both cases; however, an additional gain of chromosome 11q13 (CCND1 gene) was identified by FISH in one case. These comparative results are shown in Table 2 and also depicted in Fig. 3. Additionally, aCGH results in two melanoma samples were further studied by SNP Chip Array (Illumina); 100% and 90% of the unbalanced rearrangements detected by aCGH were confirmed by SNP array in these two cases, respectively. Despite minor discrepancies, the overall
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
4
MESBAH ARDAKANI et al.
Pathology (2017), -(-), -
(A) Low power view of H&E section of the single melanoma sample in our cohort with normal aCGH findings; this was a superficial spreading melanoma. (B) This medium power view demonstrates large areas of regression with abundant melanin pigment and melanophages together with a prominent peri-tumoural lymphocytic infiltrate resulting in low percentage of tumour cells in the microdissected area, possibly accounting for the false negative result in this case (H&E).
Fig. 1
concordance between these two platforms was excellent (Table 2).
DISCUSSION While comparative genomic hybridisation for detection of CNVs in melanocytic tumours has been in use over the past two decades, there are still many obstacles to the widespread adoption of this technology into diagnostic practice, with
testing available only in a limited number of centres worldwide.5,6,15 In particular, in a diagnostic setting which mandates utilising FFPE tissue, suboptimal DNA quality and consequent excessive background noise may effectively interfere with several steps of the experiment, such as labelling and hybridisation, therefore making data interpretation difficult. Thus, stringent optimisation, validation and quality control is required. Despite these drawbacks, in this study we
Fig. 2 Accumulated copy number changes detected in 19 melanoma samples by aCGH in the present study. In the genome view, blue indicates copy number gain and
red represents copy number loss.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
CDKN2A (9p21)
MYB (6q23)
RREB1 (6p25)
RREB1 (6p25) CCND1 (11q13)
9p21, 9q, 10p, 10q, 11q, 16q
5p, 6q23 (back ground loss of 6q17-6q25), 9p
1q
4p, 4q, 5q, 9p, 9q, 13p, 13q, 17p, 17q, 19q
3p, 4p, 4q, 9p, 9q, 10p, 10q, 11q, 12q, 14q, 15q, 17p
Loss Gain Loss
Nodular 4
1q, 3p Nodular 3
1q, 2p, 2q, 3p, 4p, 4q, 5p, 6p25 (background gain of 6p21-6p25), 7p, 7q, 8q, 12q, 16p, 16q, 17q, 18p, 18q, 19p, 19q, 20p, 20q, 21q Sub-ungal 2
1q, 2q, 6p25 Nodular 1
Gain
aCGH, array-based comparative genomic hybridisation; FISH, fluorescence in situ hybridisation.
4p, 4q, 5q, 9p, 9q, 13p, 13q, 17p, 17q, 19q, 20q
3p, 8p, 9p, 9q, 10p, 10q, 11q, 12q, 15q
Loss
Gain
FISH aCGH Melanoma subtype Lesion no.
Table 2
Copy number variations detected by aCGH in four index lesions, and further validation by FISH (in the first two lesions) and Snip Chip array (in the remaining two lesions)
SNP Array
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
5
successfully utilised a 60K array CGH to separate malignant melanomas from benign naevi through the development and implementation of meticulous analytical and post-analytical standards. Our results confirm that reliable data can be generated from this array based method with potential utility as an ancillary test with high sensitivity and specificity in separating melanocytic naevi from melanoma. The common CNVs identified in our cohort of melanomas, including losses of both arms of chromosome 9 and gains of short arms of chromosome 6 and 7, are consistent with previous reports.5,10 The high rate of chromosome 9p deletions encompassing the 9p21 region containing CDKN2A gene is of interest, supporting inclusion of a 9p21 probe in melanocytic FISH panels. CDKN2A is a tumour suppressor gene (TSG), therefore a homozygous deletion or bi-allelic inactivation of the gene by other means, resulting in complete loss of protein function, is important in melanomagenesis. It has been suggested that in a proportion of melanoma samples homozygous deletion of 9p21 can be erroneously detected as hemizygous by aCGH due to tumour heterogeneity and inadequate analytical sensitivity of the assay.10 Therefore, it is prudent to investigate potential homozygous deletions in such cases by immunohistochemical and FISH studies. A case of acral melanoma in our cohort showed multiple gains/amplifications without significant deletions, which is in keeping with the previous reports of the high load of genomic amplifications in acral melanomas.16 In a previous study, Curtin et al. found amplifications of genomic material in 89% and 85% of acral (n = 36) and mucosal (n = 20) melanomas, respectively; however, these were infrequent in melanomas from chronically or intermittently sun damaged skin.17 Similarly, our single case of lentigo maligna melanoma showed characteristic aberrations previously demonstrated in melanomas of this kind, including losses of chromosome 17p and 13q.6 Of interest, gains of chromosome 15q and losses of 17p together with overall more frequent amplifications have been reported to be common in melanomas in chronically sun damaged (CSD) skin compared to melanomas of non-CSD skin.6 Sample purity is a challenging technical difficulty in thin melanomas and can potentially lead to a false negative result. An analytical sensitivity of as low as 33% was established in our study, indicating that in a given sample a population of at least 33% tumour cells harbouring a genomic variant could be sufficient to achieve an accurate positive result. Therefore, it is necessary to maximise the tumour cell content by punctilious microdissection. On the other hand, tumour heterogeneity for a specific variant could lead to false negative results. In one melanoma sample we were not able to detect any pathogenic CNVs. This case was a superficial spreading melanoma on the scalp of an elderly man. On further histological review of the section, this case had a Breslow thickness of 1.5 mm with extensive areas of regression containing large numbers of lymphocytes and sheets of melanophages admixed with the tumour cells (Fig. 1). It is possible that a low percentage of tumour cells in the microdissected FFPE tissue could account for this negative result. In a study by Bastian et al. five of 127 (3.8%) conventional melanomas with a negative aCGH profile were identified. In two of these cases a large component of lymphocytes was present which could potentially result in a false negative result; however, in the remaining three cases no technical explanation could be given.6 Given the analytical sensitivity, while array CGH has an overall excellent performance in distinction of benign versus
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
6
MESBAH ARDAKANI et al.
Pathology (2017), -(-), -
(A) H&E section of a nodular melanoma, outlining area for microdissection in red. (B) Higher magnification of microdissected area showing highly pleomorphic malignant melanocytes with sheet like growth (H&E). (C) Array CGH data; filled red arrow indicates gain in 6p25, open red arrow indicates loss of 9p21. (D) Interphase FISH analysis using probes for 9p21 (red) and chromosome 9 centromere (green); yellow arrow indicates cells with homozygous loss of 9p21 region in the presence of two green signals consistent with intact chromosome 9 centromeres; white arrow indicates a normal cell with two green and two red signals, in keeping with intact 9p21 locus. (E) FISH analysis using RREB specific probe (red), located at 6p25; open yellow arrow indicates cells with gain of RREB characterised by clump accumulation of red signals.
Fig. 3
malignant melanocytic lesions, it is not expected to perform well for samples with a low proportion of lesional cells containing CNVs.10 As an example, in one of our melanoma samples which was further analysed by FISH, an isolated gain of chromosome 11q13 (overlapping with CCND1 gene) was detected by FISH but missed by aCGH. Clonal heterogeneity is a common finding in melanomas and has been reported previously.18 A potential limitation of this study is that only unambiguous cases were selected. Given the good performance of the assay, further validation on a cohort of ambiguous melanocytic lesions with long term clinical follow up is necessary. While there have been limited studies attempting to address this issue, the numbers of cases (particularly with documented adverse outcome) are small and the length of follow up is often short.7,19 On the other hand, to ensure an adequate yield of tumour DNA for aCGH analysis, thick melanomas were intentionally chosen. Given that nodular melanomas are often deeply invasive, they were overrepresented in this study. While this could potentially create a bias towards aberrations specific to nodular melanomas, we did not observe such a pattern. It should be noted that spitzoid melanocytic lesions can display a limited number of copy number changes. It is
becoming increasingly clear that specific genetic changes, including chromosomal abnormalities such as fusions, translocations, and point mutations are associated with specific subsets of spitzoid melanocytic tumours. There is an ever expanding list of these abnormalities including: ALK, BRAF, RET, NTRK1, ROS and MET fusions; BAP1 hemizygous deletions; HRAS mutations with or without gain of chromosome 11p; and very rarely an isolated gain of chromosome 7q.19–26 Interestingly a single case of a spitzoid neoplasm with typical histological and clinical features of a Spitz naevus has been reported showing coexisting gains of chromosome 7q and 11p.19 However this particular case and other previously reported Spitzoid neoplasms with 7q abnormality have not been further tested for other potential molecular changes such as HRAS mutations or BRAF/MET fusions.9,19 These genomic changes can occur as an isolated abnormality, possibly representing an early event in a multistep transformation,27 however they are not necessarily indicative of malignancy in isolation and usually require further genomic abnormalities to achieve malignant potential.25 This concept is similar to somatic BRAF mutation as an early event in melanomagenesis and present in both benign naevi and malignant melanomas.28 Therefore one should be
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
aware of these potential isolated changes/variants and a meticulous bio-analytical algorithm should be in place to effectively identify and interpret these lesions. For management purposes, comprehensive correlation with clinical and morphological features as well as other molecular alterations is essential to guide adequate treatment. To this end, it has been shown that fairly specific CNVs can be identified in melanomas of special type. For example Costa et al. found that melanomas associated with blue naevus/melanomas mimicking a cellular blue naevus (n = 11) demonstrate common aberrations which are similar to those previously reported in uveal melanoma.29 These CNVs included gains/amplifications of chromosome 8q, 6p and 21q and deletions/losses of chromosome 3p, 6q, 1p, 4q, 16q and 17q. Knowing these data would be of great aid in identifying the subtype of melanoma when dealing with a metastatic tumour with ambiguous histological features. It is also prudent to interpret aCGH data along with the clinical and histopathological findings and subtype of melanoma. Finally, some recent data suggest that aCGH findings can be used as a prognostic tool. It has been shown that monosomy of chromosome 3p along with deletions of 1p, 6q and 8p and duplication of 8q correlate with an overall poor prognosis in melanomas associated with/mimicking blue naevi as well as uveal melanomas.29–31 In addition Hirsch et al. discovered that fatal melanomas tend to show a significantly higher number of small focal chromosomal aberrations compared to melanomas with good prognosis and of similar clinical stage on long term follow up.32 They also described the phenomenon of ‘chromothripsis’ in melanomas, in which a single cataclysmic cellular event shatters a single or few chromosomes or small parts of a chromosome to hundreds of pieces with eventual imbalanced rearrangements resulting in numerous focal and small CNVs. In summary, we demonstrated that optimisation and application of aCGH to FFPE samples of melanocytic lesions is possible and that aCGH is a useful adjunct test in the separation of melanomas from melanocytic naevi. It should be noted that occasional naevi may show small isolated CNVs based on their biological nature. Some melanomas may not show detectable CNVs due to low tumour cell content on FFPE sections or tumour heterogeneity. It is emphasised that aCGH has not been designed to detect reciprocal or balanced rearrangements, or those involving the sex chromosomes when sex-mismatched controls are used. When used in correlation with the clinical and morphological features aCGH may prove to be a valuable adjunctive test in the interpretation of diagnostically challenging melanocytic lesions. Conflicts of interest and sources of funding: The authors state that there are no conflicts of interest to disclose. Address for correspondence: Dr Nima Mesbah Ardakani, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Hospital Avenue, Nedlands, WA 6009, Australia. E-mail: [email protected]. gov.au
References 1. Corona R, Mele A, Amini M, et al. Interobserver variability on the histopathologic diagnosis of cutaneous melanoma and other pigmented skin lesions. J Clin Oncol 1996; 14: 1218–23. 2. Farmer ER, Gonin R, Hanna MP. Discordance in the histopathologic diagnosis of melanoma and melanocytic nevi between expert pathologists. Hum Pathol 1996; 27: 528–31.
7
3. McGinnis KS, Lessin SR, Elder DE, et al. Pathology review of cases presenting to a multidisciplinary pigmented lesion clinic. Arch Dermatol 2002; 138: 617–21. 4. Troxel DB. Trends in pathology malpractice claims. Am J Surg Pathol 2012; 36: e1–5. 5. Bastian BC, LeBoit PE, Hamm H, et al. Chromosomal gains and losses in primary cutaneous melanomas detected by comparative genomic hybridization. Cancer Res 1998; 58: 2170–5. 6. Bastian BC, Olshen AB, LeBoit PE, et al. Classifying melanocytic tumors based on DNA copy number changes. Am J Pathol 2003; 163: 1765–70. 7. Raskin L, Ludgate M, Iyer RK, et al. Copy number variations and clinical outcome in atypical spitz tumors. Am J Surg Pathol 2011; 35: 243–52. 8. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature 2013; 500: 415–21. 9. Gerami P, Jewell SS, Morrison LE, et al. Fluorescence in situ hybridization (FISH) as an ancillary diagnostic tool in the diagnosis of melanoma. Am J Surg Pathol 2009; 33: 1146–56. 10. Wang L, Rao M, Fang Y, et al. A genome-wide high-resolution arrayCGH analysis of cutaneous melanoma and comparison of array-CGH to FISH in diagnostic evaluation. J Mol Diagn 2013; 15: 581–91. 11. Gerami P, Zembowicz A. Update on fluorescence in situ hybridization in melanoma: state of the art. Arch Pathol Lab Med 2011; 135: 830–7. 12. Gammon B, Beilfuss B, Guitart J, et al. Enhanced detection of spitzoid melanomas using fluorescence in situ hybridization with 9p21 as an adjunctive probe. Am J Surg Pathol 2012; 36: 81–8. 13. MacDonald JR, Ziman R, Yuen RK, et al. The Database of Genomic Variants: a curated collection of structural variation in the human genome. Nucleic Acids Res 2014; 42: D986–92. 14. Cancer Genome Atlas Network. Genomic classification of cutaneous melanoma. Cell 2015; 161: 1681–96. 15. Shlien A, Malkin D. Copy number variations and cancer. Genome Med 2009; 1: 62. 16. Bastian BC, Kashani-Sabet M, Hamm H, et al. Gene amplifications characterize acral melanoma and permit the detection of occult tumor cells in the surrounding skin. Cancer Res 2000; 60: 1968–73. 17. Curtin JA, Fridlyand J, Kageshita T, et al. Distinct sets of genetic alterations in melanoma. N Engl J Med 2005; 353: 2135–47. 18. Bogdan I, Xin H, Burg G, et al. Heterogeneity of allelic deletions within melanoma metastases. Melanoma Res 2001; 11: 349–54. 19. Ali L, Helm T, Cheney R, et al. Correlating array comparative genomic hybridization findings with histology and outcome in spitzoid melanocytic neoplasms. Int J Clin Exp Pathol 2010; 3: 593. 20. van Engen-van Grunsven AC, Kusters-Vandevelde H, Groenen PJ, et al. Update on molecular pathology of cutaneous melanocytic lesions: What is new in diagnosis and molecular testing for treatment? Front Med 2014; 1: 39. 21. Van Engen-van Grunsven A, van Dijk MC, Ruiter DJ, et al. HRASmutated Spitz tumors: a subtype of Spitz tumors with distinct features. Am J Surg Pathol 2010; 34: 1436–41. 22. Yeh I, Botton T, Talevich E, et al. Activating MET kinase rearrangements in melanoma and Spitz tumours. Nat Commun 2015; 6: 7174. 23. Busam KJ, Kutzner H, Cerroni L, et al. Clinical and pathologic findings of Spitz nevi and atypical spitz tumors with ALK fusions. Am J Surg Pathol 2014; 38: 925–33. 24. Wiesner T, He J, Yelensky R, et al. Kinase fusions are frequent in Spitz tumours and spitzoid melanomas. Nat Commun 2014; 5: 3116. 25. Yeh I, de la Fouchardiere A, Pissaloux D, et al. Clinical, histopathologic, and genomic features of spitz tumors with ALK fusions. Am J Surg Pathol 2015; 39: 581–91. 26. Botton T, Yeh I, Nelson T, et al. Recurrent BRAF kinase fusions in melanocytic tumors offer an opportunity for targeted therapy. Pigment Cell Melanoma Res 2013; 26: 845–51. 27. Shain AH, Yeh I, Kovalyshyn I, et al. The genetic evolution of melanoma from precursor lesions. N Engl J Med 2015; 373: 1926–36. 28. Yancovitz M, Litterman A, Yoon J, et al. Intra-and inter-tumor heterogeneity of BRAF (V600E) mutations in primary and metastatic melanoma. PloS One 2012; 7: e29336. 29. Costa S, Byrne M, Pissaloux D, et al. Melanomas associated with blue nevi or mimicking cellular blue nevi: clinical, pathologic, and molecular study of 11 cases displaying a high frequency of GNA11 mutations, BAP1 expression loss, and a predilection for the scalp. Am J Surg Pathol 2016; 40: 368–77. 30. Cassoux N, Rodrigues MJ, Plancher C, et al. Genome-wide profiling is a clinically relevant and affordable prognostic test in posterior uveal melanoma. Br J Ophthalmol 2014; 98: 769–74. 31. Aalto Y, Eriksson L, Seregard S, et al. Concomitant loss of chromosome 3 and whole arm losses and gains of chromosome 1, 6, or 8 in metastasizing primary uveal melanoma. Invest Ophthalmol Vis Sci 2001; 42: 313–7. 32. Hirsch D, Kemmerling R, Davis S, et al. Chromothripsis and focal copy number alterations determine poor outcome in malignant melanoma. Cancer Res 2013; 73: 1454–60.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
A N ATO M I C A L PAT H O L O G Y
Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation NIMA MESBAH ARDAKANI1,2, CARLA THOMAS1, CLEO ROBINSON1,2, KYM MINA2,3, NATHAN TOBIAS HARVEY1,2, BENHUR AMANUEL1,2 AND BENJAMIN ANDREW WOOD1,2 1
Department of Anatomical Pathology, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, 2University of Western Australia, School of Pathology and Laboratory Medicine, Crawley, and 3Department of Diagnostic Genomics, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Nedlands, WA, Australia
Summary Distinction between melanocytic naevi and melanoma occasionally poses a diagnostic challenge in ambiguous cases showing overlapping histological features. Melanomas are characterised by the presence of multiple genomic copy number variants (CNVs), while this is not a feature of naevi. We assessed the feasibility and utility of array-based comparative genomic hybridisation (aCGH) to assess CNVs in melanocytic lesions. DNA was extracted from formalin fixed, paraffin embedded (FFPE) sections of unambiguous naevi (n = 19) and melanomas (n = 19). The test DNA and gender mismatched human reference DNA were differentially labelled with fluorophores. Equal quantities of the two DNA samples were mixed and cohybridised to a SurePrint G3 Human CGH 8x60K array, and digitally scanned to capture and quantify the relative fluorescence intensities. The ratio of the fluorescence intensities was analysed by Cytogenomics software (Agilent). Frequent large CNVs were identified in 94.7% of melanoma samples, including losses of 9p (73.6%), 9q (52.6%), 10q (36.8%), 11q (36.8%), 3p (21%), and 10p (21%), and gains of 6p (42.1%), 7p (42.1%), 1q (36.8%), 8q (31.5%) and 20q (21%). Only one naevus showed two large copy number changes. Overall aCGH showed a specificity and sensitivity of 94.7% in separating naevi from melanomas. Based on our results, aCGH can be successfully used to analyse CNVs of melanocytic lesions utilising FFPE derived biopsy samples, providing a potentially useful adjunctive test for the classification of diagnostically challenging melanocytic proliferations. Key words: Comparative genomic hybridisation; copy number variations; melanoma; melanocytic naevi; virtual karyotyping. Received 26 June, revised 24 October, accepted 7 November 2016 Available online: xxx
INTRODUCTION Distinction of melanocytic naevi from cutaneous melanoma by histopathological examination is a common component of the everyday workload for many pathologists. While
morphological criteria for this separation are relatively well established, in practice these are not always easily applicable, and there are occasional ambiguous lesions characterised by overlapping features. This is reflected in several studies demonstrating only moderate interobserver agreement for the diagnosis of melanocytic lesions1,2 while others have reported that up to 11% of melanocytic diagnoses are significantly changed after specialist review.3 Misdiagnosis can lead to significant under- or overtreatment, and a false negative diagnosis of cutaneous melanoma remains the most common reason for malpractice claims against pathologists.4 Thus, there exists a clear need for ancillary testing to aid in the distinction between benign and malignant melanocytic proliferations in the setting of ambiguous histological features. The potential for cytogenetic aberrations to provide a diagnostic tool for separating melanoma from benign proliferations has been the subject of previous investigations.5,6 Gains and losses of genetic material are common in melanomas, but (with the exception of specific single abnormalities in Spitz naevi) not in melanocytic naevi.7 Indeed melanoma is notorious for showing the most unstable genome amongst all human solid tumours.8 Customised fluorescence in situ hybridisation (FISH) probes have been developed based on common chromosomal changes described in melanomas; however, these probes only assess limited loci and cannot interrogate the whole genome.9 Array-based comparative genomic hybridisation (aCGH) can detect chromosomal gains and losses of variable size across the entire genome.10 Despite the attractiveness of aCGH as a diagnostic tool, its use in clinical practice is extremely limited, in part due to significant technical requirements in regard to assay optimisation and validation for formalin fixed, paraffin embedded (FFPE) tissue and the need for microdissection to enable isolation of relatively pure lesional DNA.11 We assessed the feasibility and utility of aCGH in detection of CNVs in a cohort of unambiguous melanocytic lesions.
MATERIALS AND METHODS Selection of cases This study was approved as part of an intradepartmental routine quality and development exercise. A total of 45 samples of FFPE tissue reported as
Print ISSN 0031-3025/Online ISSN 1465-3931 Crown Copyright © 2017 Published by Elsevier B.V. on behalf of Royal College of Pathologists of Australasia. All rights reserved. DOI: http://dx.doi.org/10.1016/j.pathol.2016.11.008 Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
2
Pathology (2017), -(-), -
MESBAH ARDAKANI et al.
unambiguous malignant melanoma (MM) or melanocytic naevus were identified from the archives of Department of Anatomical Pathology, PathWest Laboratory Medicine (QEII Medical Centre, Perth, Western Australia). All cases were reviewed by two dermatopathologists (NMA and BAW) to confirm the original classification based on routine histological criteria. Two haematoxylin and eosin (H&E) stained slides representative of tumour blocks, one before and the other after 15 unstained slides, were examined and marked to select areas of viable tumour with maximum lesional cell content. These were then micro-dissected by scraping off 5 mm sections on glass slides using a scalpel blade. To be included in this study, lesions were required to have a minimum tumour thickness of 1 mm (using the Breslow methodology). Normal male and female genomic DNA (Promega, Australia) was used as reference DNA for array CGH experiments. Genomic DNA extraction Genomic DNA was extracted from FFPE tissue using the Qiagen DNeasy Blood and Tissue Kit, with modifications recommended by the ULS labelling system manufacturer (Agilent, Australia). Briefly, approximately 15 microdissected sections were heat deparaffinised at 90 C, followed by overnight treatment with 1 M sodium thiocyanate. This was followed by 48 h proteinase K treatment and then RNase A treatment. DNA was then purified using the Qiagen DNeasy Blood and Tissue Kit (Agilent), according to the manufacturer’s instructions but substituting the wash buffer AW2 with 80% ethanol and eluting the DNA in nuclease-free water. Extracted DNA was quantified by spectrophotometry using a NanoDrop ND-2000 (NanoDrop, USA). Ratio of absorbance at 260/280 was used to assess DNA purity, and samples with a ratio of ~1.80 were regarded as sufficiently pure and suitable for ULS labelling. All DNA samples were visualised on 1.0% agarose gel and fragment sizes were assessed against a 1 kb DNA ladder. DNA labelling DNA was labelled using an optimised version of the protocol for ULS labelling of FFPE DNA (Agilent). Prior to labelling, reference and FFPE DNA was heat fragmented at 95 C for 10 and 2 min, respectively, then 250 ng of tumour and reference DNA was chemically labelled by incubating with 0.5 mL of ULS-Cy3 and Cy5, respectively, in a thermal cycler with a heated lid for 30 min. Un-reacted dye was removed using KREApure filters (Agilent). The degree of labelling (DoL) was determined according to the manufacturer’s recommendations using the NanoDrop ND-2000. DoL values between 0.75% and 2.5% were regarded as optimal for Cy5, while values between 1.75% and 3.5% were optimal for Cy3-labelled DNA. Array hybridisation and scanning Cy3-labelled tumour DNA was combined with an equivalent amount of Cy5labelled sex mismatched reference DNA. Repetitive sequences were blocked with human Cot-1 DNA (Invitrogen, USA) and samples were hybridised onto SurePrint G3 Human CGH Microarrays, 8x60K (Agilent) according to manufacturer’s instructions. Following hybridisation for 40 h, microarray slides were washed according to manufacturer’s instructions and stored under nitrogen before being scanned on a DNA Microarray Scanner (Agilent). Data analysis Scanned images were analysed using Feature Extraction software (Agilent), which normalises the fluorescent intensity of both dyes at each probe and calculates their ratio, expressed on a logarithmic scale (probe log2 ratio). It also computes a set of quality control (QC) metrics, including the average green and red signal intensity at all the probes as well as the background signal (noise) and signal-to-noise ratio using non-hybridising control probes. Feature extracted data were then analysed using CytoGenomics Software (Agilent). FISH analysis Five mm thick, unstained, FFPE tissue sections were subject to in situ hybridisation with melanoma four-colour FISH kit including probes for RREB1 (6p25), CEP 6 (6p11.1-q11.1), MYB (6q23), and CCND1 (11q13) as well as CDKN2A/CEP 9 FISH probe kit (Vysis, USA).11,12 A standard hybridisation protocol was followed, as previously reported.9 A minimum of 60 tumour nuclei were counted for FISH analysis.
SNP array analysis Ten unstained sections from FFPE material were obtained for microdissection, with a single prior and subsequent H&E stained section used to confirm adequacy of tumour tissue and to mark areas for microdissection. DNA was extracted using a Qiagen extraction kit according to the manufacturer’s instruction (Qiagen, Germany). The Infinium HD assay was performed utilising Illumina iScan and HumanCytoSNP FFPE-12 BeadChip array according to the manufacturer’s protocol (Illumina, USA). The BeadChips were stained, and then imaged, using a BeadArray Reader (Illumina). Image data were analysed with GenomeStudio (Illumina).
RESULTS Bio-analytical adjustment aCGH primary data were analysed according to the Aberration Detection Method 2 (ADM-2) algorithm as previously described.10 The sensitivity threshold for the ADM-2 algorithm was adjusted based on self-self experiments performed on two samples. The post-analytic filters were also rectified based on two intra-array experiments. The optimal sensitivity threshold for the ADM-2 algorithm was defined as 6.1 after testing multiple thresholds. In addition, a minimum of 10 consecutive probes was established as the minimum postanalytic filter for CNVs which resulted in identical calls for intra-array duplicates performed on two samples. The log2 ratio threshold for detecting gains and losses of genomic material was set at 0.25. Quality control (QC) Opposite sex reference DNA was hybridised against each test DNA as an internal control for the quality and validity of results of each experiment. If the expected differences of the sex chromosomes (chromosome X and Y) were not met in a given sample, the experiment was considered unsatisfactory. QC metrics were also assessed for each sample. If QC metrics, in particular derivative log ratio (DLR) spread and red or green signal intensity, were not within the accepted limits, the experiment was considered unsatisfactory. A DLR spread of less than 0.5 was considered optimal, however DLR spread of equal or less than 0.6 was also accepted if the sex chromosome patterns were as expected. The samples with unsatisfactory results were repeated if there was adequate residual DNA or optimal remaining tissue to re-extract DNA. Post-analytic interpretation Copy number variants (CNVs) detected in each sample were first cross referenced with the Database of Genomic Variants (DGV) to exclude potential polymorphic CNVs.13 Large CNVs involving at least 100 consecutive probes were defined as pathogenic if CNVs overlapped exonic regions. CNVs involving less than 100 consecutive probes were studied individually. Variants with high log2 ratio detected by Cytogenomic software as homozygous deletions or amplifications were also considered significant if overlapped with areas on the genome harbouring genes known to be involved in melanomagenesis such as CCND1, CDKN2A, PTEN, P53, MYC, CDK4, NF1 and MDM2, or known to be frequently altered in melanomas such as MYB1 and RREB1.14 The remaining small (<100 probe) CNVs were considered as being of unknown significance.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
Analytical sensitivity of aCGH To evaluate the clonal detection threshold of the array, we serially diluted and tested a commercially available DNA sample with known germline heterozygous deletion of 13q13.3q21.33 (DNA sample NA01484; Coriell Institute, USA). Given the minimum average log2 ratio per probe was set at 0.25, the aCGH was expected to detect CNVs present in at least 33% of the constituent cells in a sample. Thus, we diluted the reference DNA by adding portions of normal DNA to achieve samples with 100%, 50%, 40%, 33%, 25% and 20% of abnormal DNA. The aCGH was able to detect the known deletion in samples with 100%, 50%, 40% and 33%, but not in samples with 25% and 20% of abnormal cells. Therefore the analytical sensitivity of the assay was established at expected 33% cut-off. Final study cohort After assessment of QC metrics and sex chromosome changes in each run, 12 samples showed technically unsatisfactory results. In five cases the quantity of remaining DNA was low and an attempt to re-extract DNA was unsuccessful due to low tumor cell content. Seven samples were repeated, of which five achieved satisfactory QC metrics on the repeat. The remaining cases were excluded from the study. After excluding the technically unsatisfactory results, 19 melanomas and 19 naevi were included in this study. The final cohort was composed of 38 lesions from 38 patients, including 24 men and 14 women with a median age of 58 years (range 15–86, mean 54.5 years). The patients in the melanoma cohort were older than the patients with benign naevi (mean age of 65.5 and 38.2 years, respectively, p < 0.05). The index lesions were thicker in the melanoma cohort with an average thickness of 6.65 mm as opposed to a mean thickness of 2.74 mm in naevi (p < 0.05). The melanoma subtypes included nine nodular, four superficial spreading, two acral lentiginous melanoma, one lentigo maligna melanoma and three metastatic melanomas. In metastatic melanomas the primary tumour was a superficial spreading melanoma in two cases, and in one case the primary tumour was unknown. In primary melanomas the Breslow thickness ranged from 1.1 mm to 15 mm with an average of 4.33 mm and a median of 3.5 mm. The melanomas were mainly located on the trunk and extremities while the naevi where predominantly from the head and neck areas. Histologically the naevi showed features of common compound or intradermal naevi which could be further classified as ‘Miescher’ naevus (n = 9), ‘Unna’ naevus (n = 3) and naevi with attributes of congenital naevus (n = 7) based on the pattern of growth. Detailed demographics are presented in Table 1. Genomic variants All but one case of melanoma demonstrated multiple large unbalanced CNVs. All but one of the melanocytic naevi showed either no CNVs or occasional small variants of unknown significance. One naevus showed two relatively large genomic losses on the long arms of chromosome 10 and 19, which were considered significant CNVs. One melanoma showed a normal aCGH profile. This case was a superficial spreading melanoma arising on the scalp of an elderly man with large areas of regression and a prominent lymphocytic
3
Table 1 Demographics of the patients and characteristics of the lesions in the study cohort Melanoma n = 19 Gender, n Male Female Age, years Range Average Thickness/depth of lesion, mm Range Average Median Melanoma subtype Superficial spreading Nodular Lentigo maligna Acral Other Site Head and neck Trunk Extremities
Naevi n = 19
12 7
12 7
23–86 65.5
15–78 38.2
1.55–18 6.65 5
1–5 2.74 2.75
4 9 1 2 3 3 9 7
10 8 1
infiltrate (Fig. 1). The sensitivity and specificity of aCGH in separation of unambiguous benign naevi from melanomas was 94.7% [confidence interval (CI) 71.8–99.7%]. The melanomas showed an average of 11.3 CNVs (range 0–25) with a mean number of 5.3 gains (range 0–22) and 6 losses (range 0–17). In most cases the losses were more numerous than gains; however, in four cases gains were predominant. One of these latter cases was an acral subungal melanoma with 22 gains and 3 losses. The most common CNVs in melanomas encompassed large losses of chromosome 9p (73.6%), 9q (52.6%), 10q (36.8%), 11q (36.8%), 3p (21%), and 10p (21%), and gains of chromosome 6p (42.1%), 7p (42.1%), 1q (36.8%), 8q (31.5%) and 20q (21%) (Fig. 2). These aberrations showed significant overlap with regions of the genome containing oncogenes or tumour suppressor genes known to be involved in melanomagenesis. For example, loss of chromosome 9p21 harbouring the CDKN2A (p16) gene was seen in 13 (68.4%) melanomas, four of which appeared to be homozygous deletions resulting in complete loss of tumour suppressor function. Other genes commonly involved in melanoma development such as RREB1, MYB, MYC and CDK4 were also well represented in our cohort. Further validation of CNVs The array CGH results were further validated by a four colour FISH assay (Vysis) with addition of a 9p21 (containing CDKN2A, p16 gene) probe (Vysis) in two melanoma cases. All aCGH calls within the detection range of the FISH assay were confirmed in both cases; however, an additional gain of chromosome 11q13 (CCND1 gene) was identified by FISH in one case. These comparative results are shown in Table 2 and also depicted in Fig. 3. Additionally, aCGH results in two melanoma samples were further studied by SNP Chip Array (Illumina); 100% and 90% of the unbalanced rearrangements detected by aCGH were confirmed by SNP array in these two cases, respectively. Despite minor discrepancies, the overall
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
4
MESBAH ARDAKANI et al.
Pathology (2017), -(-), -
(A) Low power view of H&E section of the single melanoma sample in our cohort with normal aCGH findings; this was a superficial spreading melanoma. (B) This medium power view demonstrates large areas of regression with abundant melanin pigment and melanophages together with a prominent peri-tumoural lymphocytic infiltrate resulting in low percentage of tumour cells in the microdissected area, possibly accounting for the false negative result in this case (H&E).
Fig. 1
concordance between these two platforms was excellent (Table 2).
DISCUSSION While comparative genomic hybridisation for detection of CNVs in melanocytic tumours has been in use over the past two decades, there are still many obstacles to the widespread adoption of this technology into diagnostic practice, with
testing available only in a limited number of centres worldwide.5,6,15 In particular, in a diagnostic setting which mandates utilising FFPE tissue, suboptimal DNA quality and consequent excessive background noise may effectively interfere with several steps of the experiment, such as labelling and hybridisation, therefore making data interpretation difficult. Thus, stringent optimisation, validation and quality control is required. Despite these drawbacks, in this study we
Fig. 2 Accumulated copy number changes detected in 19 melanoma samples by aCGH in the present study. In the genome view, blue indicates copy number gain and
red represents copy number loss.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
CDKN2A (9p21)
MYB (6q23)
RREB1 (6p25)
RREB1 (6p25) CCND1 (11q13)
9p21, 9q, 10p, 10q, 11q, 16q
5p, 6q23 (back ground loss of 6q17-6q25), 9p
1q
4p, 4q, 5q, 9p, 9q, 13p, 13q, 17p, 17q, 19q
3p, 4p, 4q, 9p, 9q, 10p, 10q, 11q, 12q, 14q, 15q, 17p
Loss Gain Loss
Nodular 4
1q, 3p Nodular 3
1q, 2p, 2q, 3p, 4p, 4q, 5p, 6p25 (background gain of 6p21-6p25), 7p, 7q, 8q, 12q, 16p, 16q, 17q, 18p, 18q, 19p, 19q, 20p, 20q, 21q Sub-ungal 2
1q, 2q, 6p25 Nodular 1
Gain
aCGH, array-based comparative genomic hybridisation; FISH, fluorescence in situ hybridisation.
4p, 4q, 5q, 9p, 9q, 13p, 13q, 17p, 17q, 19q, 20q
3p, 8p, 9p, 9q, 10p, 10q, 11q, 12q, 15q
Loss
Gain
FISH aCGH Melanoma subtype Lesion no.
Table 2
Copy number variations detected by aCGH in four index lesions, and further validation by FISH (in the first two lesions) and Snip Chip array (in the remaining two lesions)
SNP Array
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
5
successfully utilised a 60K array CGH to separate malignant melanomas from benign naevi through the development and implementation of meticulous analytical and post-analytical standards. Our results confirm that reliable data can be generated from this array based method with potential utility as an ancillary test with high sensitivity and specificity in separating melanocytic naevi from melanoma. The common CNVs identified in our cohort of melanomas, including losses of both arms of chromosome 9 and gains of short arms of chromosome 6 and 7, are consistent with previous reports.5,10 The high rate of chromosome 9p deletions encompassing the 9p21 region containing CDKN2A gene is of interest, supporting inclusion of a 9p21 probe in melanocytic FISH panels. CDKN2A is a tumour suppressor gene (TSG), therefore a homozygous deletion or bi-allelic inactivation of the gene by other means, resulting in complete loss of protein function, is important in melanomagenesis. It has been suggested that in a proportion of melanoma samples homozygous deletion of 9p21 can be erroneously detected as hemizygous by aCGH due to tumour heterogeneity and inadequate analytical sensitivity of the assay.10 Therefore, it is prudent to investigate potential homozygous deletions in such cases by immunohistochemical and FISH studies. A case of acral melanoma in our cohort showed multiple gains/amplifications without significant deletions, which is in keeping with the previous reports of the high load of genomic amplifications in acral melanomas.16 In a previous study, Curtin et al. found amplifications of genomic material in 89% and 85% of acral (n = 36) and mucosal (n = 20) melanomas, respectively; however, these were infrequent in melanomas from chronically or intermittently sun damaged skin.17 Similarly, our single case of lentigo maligna melanoma showed characteristic aberrations previously demonstrated in melanomas of this kind, including losses of chromosome 17p and 13q.6 Of interest, gains of chromosome 15q and losses of 17p together with overall more frequent amplifications have been reported to be common in melanomas in chronically sun damaged (CSD) skin compared to melanomas of non-CSD skin.6 Sample purity is a challenging technical difficulty in thin melanomas and can potentially lead to a false negative result. An analytical sensitivity of as low as 33% was established in our study, indicating that in a given sample a population of at least 33% tumour cells harbouring a genomic variant could be sufficient to achieve an accurate positive result. Therefore, it is necessary to maximise the tumour cell content by punctilious microdissection. On the other hand, tumour heterogeneity for a specific variant could lead to false negative results. In one melanoma sample we were not able to detect any pathogenic CNVs. This case was a superficial spreading melanoma on the scalp of an elderly man. On further histological review of the section, this case had a Breslow thickness of 1.5 mm with extensive areas of regression containing large numbers of lymphocytes and sheets of melanophages admixed with the tumour cells (Fig. 1). It is possible that a low percentage of tumour cells in the microdissected FFPE tissue could account for this negative result. In a study by Bastian et al. five of 127 (3.8%) conventional melanomas with a negative aCGH profile were identified. In two of these cases a large component of lymphocytes was present which could potentially result in a false negative result; however, in the remaining three cases no technical explanation could be given.6 Given the analytical sensitivity, while array CGH has an overall excellent performance in distinction of benign versus
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
6
MESBAH ARDAKANI et al.
Pathology (2017), -(-), -
(A) H&E section of a nodular melanoma, outlining area for microdissection in red. (B) Higher magnification of microdissected area showing highly pleomorphic malignant melanocytes with sheet like growth (H&E). (C) Array CGH data; filled red arrow indicates gain in 6p25, open red arrow indicates loss of 9p21. (D) Interphase FISH analysis using probes for 9p21 (red) and chromosome 9 centromere (green); yellow arrow indicates cells with homozygous loss of 9p21 region in the presence of two green signals consistent with intact chromosome 9 centromeres; white arrow indicates a normal cell with two green and two red signals, in keeping with intact 9p21 locus. (E) FISH analysis using RREB specific probe (red), located at 6p25; open yellow arrow indicates cells with gain of RREB characterised by clump accumulation of red signals.
Fig. 3
malignant melanocytic lesions, it is not expected to perform well for samples with a low proportion of lesional cells containing CNVs.10 As an example, in one of our melanoma samples which was further analysed by FISH, an isolated gain of chromosome 11q13 (overlapping with CCND1 gene) was detected by FISH but missed by aCGH. Clonal heterogeneity is a common finding in melanomas and has been reported previously.18 A potential limitation of this study is that only unambiguous cases were selected. Given the good performance of the assay, further validation on a cohort of ambiguous melanocytic lesions with long term clinical follow up is necessary. While there have been limited studies attempting to address this issue, the numbers of cases (particularly with documented adverse outcome) are small and the length of follow up is often short.7,19 On the other hand, to ensure an adequate yield of tumour DNA for aCGH analysis, thick melanomas were intentionally chosen. Given that nodular melanomas are often deeply invasive, they were overrepresented in this study. While this could potentially create a bias towards aberrations specific to nodular melanomas, we did not observe such a pattern. It should be noted that spitzoid melanocytic lesions can display a limited number of copy number changes. It is
becoming increasingly clear that specific genetic changes, including chromosomal abnormalities such as fusions, translocations, and point mutations are associated with specific subsets of spitzoid melanocytic tumours. There is an ever expanding list of these abnormalities including: ALK, BRAF, RET, NTRK1, ROS and MET fusions; BAP1 hemizygous deletions; HRAS mutations with or without gain of chromosome 11p; and very rarely an isolated gain of chromosome 7q.19–26 Interestingly a single case of a spitzoid neoplasm with typical histological and clinical features of a Spitz naevus has been reported showing coexisting gains of chromosome 7q and 11p.19 However this particular case and other previously reported Spitzoid neoplasms with 7q abnormality have not been further tested for other potential molecular changes such as HRAS mutations or BRAF/MET fusions.9,19 These genomic changes can occur as an isolated abnormality, possibly representing an early event in a multistep transformation,27 however they are not necessarily indicative of malignancy in isolation and usually require further genomic abnormalities to achieve malignant potential.25 This concept is similar to somatic BRAF mutation as an early event in melanomagenesis and present in both benign naevi and malignant melanomas.28 Therefore one should be
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008
COMPARATIVE GENOMIC HYBRIDISATION IN MELANOCYTIC LESIONS
aware of these potential isolated changes/variants and a meticulous bio-analytical algorithm should be in place to effectively identify and interpret these lesions. For management purposes, comprehensive correlation with clinical and morphological features as well as other molecular alterations is essential to guide adequate treatment. To this end, it has been shown that fairly specific CNVs can be identified in melanomas of special type. For example Costa et al. found that melanomas associated with blue naevus/melanomas mimicking a cellular blue naevus (n = 11) demonstrate common aberrations which are similar to those previously reported in uveal melanoma.29 These CNVs included gains/amplifications of chromosome 8q, 6p and 21q and deletions/losses of chromosome 3p, 6q, 1p, 4q, 16q and 17q. Knowing these data would be of great aid in identifying the subtype of melanoma when dealing with a metastatic tumour with ambiguous histological features. It is also prudent to interpret aCGH data along with the clinical and histopathological findings and subtype of melanoma. Finally, some recent data suggest that aCGH findings can be used as a prognostic tool. It has been shown that monosomy of chromosome 3p along with deletions of 1p, 6q and 8p and duplication of 8q correlate with an overall poor prognosis in melanomas associated with/mimicking blue naevi as well as uveal melanomas.29–31 In addition Hirsch et al. discovered that fatal melanomas tend to show a significantly higher number of small focal chromosomal aberrations compared to melanomas with good prognosis and of similar clinical stage on long term follow up.32 They also described the phenomenon of ‘chromothripsis’ in melanomas, in which a single cataclysmic cellular event shatters a single or few chromosomes or small parts of a chromosome to hundreds of pieces with eventual imbalanced rearrangements resulting in numerous focal and small CNVs. In summary, we demonstrated that optimisation and application of aCGH to FFPE samples of melanocytic lesions is possible and that aCGH is a useful adjunct test in the separation of melanomas from melanocytic naevi. It should be noted that occasional naevi may show small isolated CNVs based on their biological nature. Some melanomas may not show detectable CNVs due to low tumour cell content on FFPE sections or tumour heterogeneity. It is emphasised that aCGH has not been designed to detect reciprocal or balanced rearrangements, or those involving the sex chromosomes when sex-mismatched controls are used. When used in correlation with the clinical and morphological features aCGH may prove to be a valuable adjunctive test in the interpretation of diagnostically challenging melanocytic lesions. Conflicts of interest and sources of funding: The authors state that there are no conflicts of interest to disclose. Address for correspondence: Dr Nima Mesbah Ardakani, PathWest Laboratory Medicine, Queen Elizabeth II Medical Centre, Hospital Avenue, Nedlands, WA 6009, Australia. E-mail: [email protected]. gov.au
References 1. Corona R, Mele A, Amini M, et al. Interobserver variability on the histopathologic diagnosis of cutaneous melanoma and other pigmented skin lesions. J Clin Oncol 1996; 14: 1218–23. 2. Farmer ER, Gonin R, Hanna MP. Discordance in the histopathologic diagnosis of melanoma and melanocytic nevi between expert pathologists. Hum Pathol 1996; 27: 528–31.
7
3. McGinnis KS, Lessin SR, Elder DE, et al. Pathology review of cases presenting to a multidisciplinary pigmented lesion clinic. Arch Dermatol 2002; 138: 617–21. 4. Troxel DB. Trends in pathology malpractice claims. Am J Surg Pathol 2012; 36: e1–5. 5. Bastian BC, LeBoit PE, Hamm H, et al. Chromosomal gains and losses in primary cutaneous melanomas detected by comparative genomic hybridization. Cancer Res 1998; 58: 2170–5. 6. Bastian BC, Olshen AB, LeBoit PE, et al. Classifying melanocytic tumors based on DNA copy number changes. Am J Pathol 2003; 163: 1765–70. 7. Raskin L, Ludgate M, Iyer RK, et al. Copy number variations and clinical outcome in atypical spitz tumors. Am J Surg Pathol 2011; 35: 243–52. 8. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature 2013; 500: 415–21. 9. Gerami P, Jewell SS, Morrison LE, et al. Fluorescence in situ hybridization (FISH) as an ancillary diagnostic tool in the diagnosis of melanoma. Am J Surg Pathol 2009; 33: 1146–56. 10. Wang L, Rao M, Fang Y, et al. A genome-wide high-resolution arrayCGH analysis of cutaneous melanoma and comparison of array-CGH to FISH in diagnostic evaluation. J Mol Diagn 2013; 15: 581–91. 11. Gerami P, Zembowicz A. Update on fluorescence in situ hybridization in melanoma: state of the art. Arch Pathol Lab Med 2011; 135: 830–7. 12. Gammon B, Beilfuss B, Guitart J, et al. Enhanced detection of spitzoid melanomas using fluorescence in situ hybridization with 9p21 as an adjunctive probe. Am J Surg Pathol 2012; 36: 81–8. 13. MacDonald JR, Ziman R, Yuen RK, et al. The Database of Genomic Variants: a curated collection of structural variation in the human genome. Nucleic Acids Res 2014; 42: D986–92. 14. Cancer Genome Atlas Network. Genomic classification of cutaneous melanoma. Cell 2015; 161: 1681–96. 15. Shlien A, Malkin D. Copy number variations and cancer. Genome Med 2009; 1: 62. 16. Bastian BC, Kashani-Sabet M, Hamm H, et al. Gene amplifications characterize acral melanoma and permit the detection of occult tumor cells in the surrounding skin. Cancer Res 2000; 60: 1968–73. 17. Curtin JA, Fridlyand J, Kageshita T, et al. Distinct sets of genetic alterations in melanoma. N Engl J Med 2005; 353: 2135–47. 18. Bogdan I, Xin H, Burg G, et al. Heterogeneity of allelic deletions within melanoma metastases. Melanoma Res 2001; 11: 349–54. 19. Ali L, Helm T, Cheney R, et al. Correlating array comparative genomic hybridization findings with histology and outcome in spitzoid melanocytic neoplasms. Int J Clin Exp Pathol 2010; 3: 593. 20. van Engen-van Grunsven AC, Kusters-Vandevelde H, Groenen PJ, et al. Update on molecular pathology of cutaneous melanocytic lesions: What is new in diagnosis and molecular testing for treatment? Front Med 2014; 1: 39. 21. Van Engen-van Grunsven A, van Dijk MC, Ruiter DJ, et al. HRASmutated Spitz tumors: a subtype of Spitz tumors with distinct features. Am J Surg Pathol 2010; 34: 1436–41. 22. Yeh I, Botton T, Talevich E, et al. Activating MET kinase rearrangements in melanoma and Spitz tumours. Nat Commun 2015; 6: 7174. 23. Busam KJ, Kutzner H, Cerroni L, et al. Clinical and pathologic findings of Spitz nevi and atypical spitz tumors with ALK fusions. Am J Surg Pathol 2014; 38: 925–33. 24. Wiesner T, He J, Yelensky R, et al. Kinase fusions are frequent in Spitz tumours and spitzoid melanomas. Nat Commun 2014; 5: 3116. 25. Yeh I, de la Fouchardiere A, Pissaloux D, et al. Clinical, histopathologic, and genomic features of spitz tumors with ALK fusions. Am J Surg Pathol 2015; 39: 581–91. 26. Botton T, Yeh I, Nelson T, et al. Recurrent BRAF kinase fusions in melanocytic tumors offer an opportunity for targeted therapy. Pigment Cell Melanoma Res 2013; 26: 845–51. 27. Shain AH, Yeh I, Kovalyshyn I, et al. The genetic evolution of melanoma from precursor lesions. N Engl J Med 2015; 373: 1926–36. 28. Yancovitz M, Litterman A, Yoon J, et al. Intra-and inter-tumor heterogeneity of BRAF (V600E) mutations in primary and metastatic melanoma. PloS One 2012; 7: e29336. 29. Costa S, Byrne M, Pissaloux D, et al. Melanomas associated with blue nevi or mimicking cellular blue nevi: clinical, pathologic, and molecular study of 11 cases displaying a high frequency of GNA11 mutations, BAP1 expression loss, and a predilection for the scalp. Am J Surg Pathol 2016; 40: 368–77. 30. Cassoux N, Rodrigues MJ, Plancher C, et al. Genome-wide profiling is a clinically relevant and affordable prognostic test in posterior uveal melanoma. Br J Ophthalmol 2014; 98: 769–74. 31. Aalto Y, Eriksson L, Seregard S, et al. Concomitant loss of chromosome 3 and whole arm losses and gains of chromosome 1, 6, or 8 in metastasizing primary uveal melanoma. Invest Ophthalmol Vis Sci 2001; 42: 313–7. 32. Hirsch D, Kemmerling R, Davis S, et al. Chromothripsis and focal copy number alterations determine poor outcome in malignant melanoma. Cancer Res 2013; 73: 1454–60.
Please cite this article in press as: Mesbah Ardakani N, et al., Detection of copy number variations in melanocytic lesions utilising array based comparative genomic hybridisation, Pathology (2017), http://dx.doi.org/10.1016/j.pathol.2016.11.008