Somatic mutation testing—technology

BRAF targets in melanoma. Biological mechanisms, resistance, and drug discovery. Cancer drug discovery and development. Volume 82. Ed. Ryan J. Sullivan. Springer (2015)


Understanding of the genetic underpinnings of melanoma has led to current treatment advances for advanced stage melanoma and will continue to aid in the development of future therapies. Therefore, it is important to identify known mutations in melanoma tumors in order to stratify patients for therapeutic options, as well as identify mechanisms and mutations involved in treatment resistance. A number of techniques have been used to identify somatic mutations and genomic aberrations providing clinicians with tools to genotype melanoma tumor samples from patients, at all stages of disease.

For many years, molecular diagnostic techniques have evaluated single gene mutations individually or a small number of genes through reaction multiplexing. Massively parallel sequencing allows for the simultaneous testing and identification of multiple mutations and genomic aberrations within tumor samples concurrently.

Although knowledge of all mutations and genomic aberrations within tumor samples would appear on the surface to be most helpful, currently there are limited gene mutations that are clinically actionable. Thus, assaying individual genes and/ or mutations is still appropriate in many circumstances. It is important to note that although full profiling of tumors may shed light on future research and clinical trial endeavors, it is very possible that mutations will be identified for which no therapeutic intervention is currently available.

Tumor samples are heterogeneous, which may result in only a fraction of tumor cells harboring a specific mutation, and also may contain surrounding normal tissue resulting in decreased amount of mutated DNA in tumor samples (admixture). Thus, assay sensitivity is important so that mutations can be detected even when they represent a small portion of the DNA extracted from the tumor sample. Advances in several technologies have allowed for the detection of mutations in samples with as little as 5% mutant DNA in the total DNA sample. Different sources of tumor samples are available for testing including fresh frozen tumor samples and formalin fixed paraffin-embedded (FFPE) tumor samples. FFPE tumor samples, which are commonly used for clinical mutation detection, can have DNA which is degraded and fragmented [63]. A specimen of large enough size for DNA extraction also needs to be available, which can be particularly an issue for primary melanomas. However, in the vast majority of cases, FFPE specimens from metastatic melanomas can be used for mutation identification, even for alleles at relatively low frequency.

When evaluating tumor samples with gene specific mutation testing, consideration must be given to the type of mutation being sought. Some somatic point mutations occur at specific sites in a given gene, known as hotspot mutations (which can be seen in oncogenes), whereas other mutations can occur anywhere within a gene (which can be seen in tumor suppressor genes). Mutational patterns will dictate the type of analysis optimal for mutational detection, as does the number of samples being analyzed, as some methods are better suited for processing of multiple samples, and others more appropriate for limited numbers of samples. We will review techniques used in clinical laboratories focusing on individual gene testing, along with newer sequencing technologies used to identify mutations within melanoma tumor samples.

Direct sequencing

DNA isolated from tumor samples can undergo direct sequencing to identify point mutations in a specific stretch of DNA. Sanger sequencing, or chain terminating method, can be performed on DNA from tumor samples using a variety of dyeterminators, but is relatively insensitive with a mutation detection rate of ~ 25% allele frequency [64, 65]. Pyrosequencing™ (Qiagen, Inc., Alameda, CA) is another direct sequencing technique [66] and can be used to sequence specific short regions of DNA up to 50 bases. Somatic mutations can be identified when clustered within a small region of interest providing for the identification of mutations within a given DNA locus. Pyrosequencing™ is used by many molecular pathology laboratories to evaluate somatic mutations located within mutation hotspots, and has the advantage of detecting mutant DNA alleles at frequencies as low as 5–15% of the total, depending on the gene being investigated [63, 66]. This method is useful for sequencing BRAF mutations in tumor samples, as mutations have been identified in several different nucleotides within and around BRAF V600 [32–36].

Allele-specific primers are used to detect single nucleotide changes in tumor samples. For single mutations, Taqman® mutation detection assays (Life Technologies, Carlsbad, CA) is a popular choice. Single nucleotide extension assays also can be used to identify specific point mutations in a given gene, as the technique evaluates changes at an individual nucleotide. Two commonly used platforms for multiplexed single nucleotide extension assays include iPlex™ (Sequenom, Inc, San Diego, CA) [67, 68] and SNaPshot™ (Applied Biosystems, Inc, Foster City, CA) [69]. These techniques make use of primer sets to amplify the DNA and detect the mutated base, along with specific tags, which results in amplification and multiplexing [70]; the tags vary depending upon the platform that is employed. For the iPlex™ platform, nucleotides are detected by matrix-assisted laser desorption/ ionization, time-of-flight mass spectrometry (MALDI-TOF) analysis [71]. For the SNaPshot™ platform, nucleotides are fluorescently labeled and nucleotide incorporation into extension products are detected [70].

The iPlex™ [72] and SNaPshot™ [70] technologies can detect mutant DNA with a sensitivity of 5–10% of DNA, thus demonstrating a higher sensitivity for mutation detection than direct sequencing. Additionally, these platforms are effective in genotyping DNA from FFPE tumor samples, allowing for mutation detection in lower quality DNA. Given the use of primer tags, multiple single nucleotide extension assays can be multiplexed, allowing for the interrogation of a number of different mutations within a given reaction. Multiple mutations or single nucleotide polymorphisms within a given region can be assessed in specific tumor samples, albeit not within the same multiplex. These platforms are also commonly used by molecular pathology laboratories and are well suited to assess genes which demonstrate mutational hotspots, such as NRAS, BRAF, and GNA11/GNAQ.

In 2011, the United States Food and Drug Administration (FDA) approved the targeted mutant BRAF inhibitor, vemurafenib, for the treatment of advanced melanoma in patients with the BRAF V600E mutation [3]. As this therapy is targeted to a specific somatic mutation identified in patient tumor samples, tumor samples must undergo molecular testing to detect this mutation prior to initiation of therapy. Simultaneously, with the approval of vemurafenib, the FDA also approved a commercially available test for the BRAF V600E mutation, the cobas® 4800 BRAF V600 Mutation Test, in order to determine the presence of the BRAF mutation and to receive treatment with vemurafenib. The cobas® test probes are specific for and bind to wild-type and mutant V600E BRAF sequences and are detected when the probes bind to their correct sequence. However, the cobas® test is limited in its ability to detect BRAF non-V600E mutations, with a 66% cross-sensitivity for BRAF V600K, and V600E mutations that are a result of a two base pair mutation (package insert). Evidence suggests that patients with non-V600E BRAF mutations, such as V600K, also respond to therapy with targeted BRAF inhibitors, such as vemurafenib and dabrafenib [3, 11, 12]. As such, it is important that somatic mutations in patient tumor samples are accurately detected, thus using the cobas® test alone may not be adequate.

Genomic aberrations

DNA copy number alterations have been shown to be involved in the pathogenesis of a number of cancers [73] and may have predictive value relating to disease progression or clinical outcome in different tumor types [74–78]. Copy gains or losses are structural variants of segments of DNA, and thought to exert their effects through dysregulation of gene expression. Techniques to determine cancer copy number profiles have improved over the years. Initially copy number determination was performed with probe sequences derived from bacterial artificial chromosomes. Single nucleotide polymorphism (SNP)-based arrays and array-based Comparative Genomic Hybridization (aCGH) are used currently, with massively parallel sequencing the next step for copy number determination, with the development of precise analytical techniques needed. SNP arrays also provide increased identification of loss of heterozygosity and allele copy number, however, these arrays are less suitable for analysis of FFPE tumor samples given the concerns of DNA quality. DNA degradation is common in FFPE tumor samples. As such, shorter DNA fragments are present and limit accurate detection of copy gains and losses in tumor samples. These array techniques rely on the presence of longer DNA fragment sizes to map regions of copy gain and loss, thus, these shorter fragments can result in increased background signal in assay data and could contribute to imprecise DNA copy gains and losses in tumor samples.

Copy number alterations have been analyzed in melanoma cell lines and tumor samples in order to detect genomic aberrations and distinct genomic changes involved in melanoma pathogenesis [79]. A number of genetic regions have been found to be altered in melanoma tumor samples, including gain of chromosomes 5 and 7 and loss of chromosomes 4, 10, 11, 12, 17, and 22 [35,80]. Amplifications of BRAF, NRAS, MITF, CCND1, MDM2, and NOTCH2, and homozygous deletions of CDKN2A and PTEN have been identified as driver aberrations [35, 81–83]. Moreover, specific patterns of chromosomal gains and losses have been associated with BRAF and NRAS mutation status [35, 80], suggesting that additional genetic alterations or aberrations cooperate in the pathogenesis of these melanomas.

Other techniques evaluating genomic aberrations have been used to provide supplemental information which, in different tumor types, can be used in risk stratification and prognostic implications in clinical settings. Larger genomic alterations, deletions and rearrangements (over 100,000 base pairs) can be detected using Fluorescence in Situ Hybridization (FISH). FISH is routinely used in hematologic malignancies to delineate cytogenetic characteristics with direct impact on disease stratification and treatment decisions. FISH is gaining popularity to evaluate solid malignancies as well [84, 85]. FISH is being used to detect ALK rearrangement in lung cancers, the presence of which provides the rationale for treatment with the ALK inhibitor crizotinib [86]. FISH based assays are emerging as tools to assist in the diagnosis of histologically indeterminate melanoma. Using three specific probes for RREB1, MYB, and CCND1 genes and a centromere specific control probe, Senetta et al. [87]. assessed their use in distinguishing between benign nevi and melanoma. Although specific probe patterns were established in benign nevi vs. melanoma in the validation samples, results were ambiguous in the indeterminate samples in their sample set. Hossain et al. [88] evaluated the use chromosome specific probes to categorize benign lesions vs. melanoma. Results from these studies established chromosomal abnormalities in 94% melanoma samples, 6% compound nevi, and 0% normal skin. Moreover, the most frequent abnormality was gain of chromosomal 11, along with observed gains in chromosomes 6, 7, and 20 [88]. Clinicians can use results from FISH analysis, due to these characteristic genetic events, to guide clinical decisions in the setting of indeterminate histology.

Multiplex probe ligation amplification (MLPA, MRC-Holland, Amsterdam, Netherlands) is used to perform targeted analysis in tumor samples in order to evaluate specific, localized amplifications and deletions [89, 90]. Probes are annealed adjacent to the genomic region of interest, ligated together, and amplified. Quantification and determination of copy number is determined by normalization to controls. MLPA provides copy number profiles for specific genes of interest, requiring less tumor DNA as starting material relative to aCGH. In addition, MLPA can be multiplexed to evaluate a number of genes within the same reaction. MLPA had been successfully used to identify genetic rearrangements in genes associated with inherited syndromes, contiguous gene deletion syndromes, and somatic copy number alterations [91–94]. Moreover, evaluation of specific chromosomal loci, including chr 9p21 (?CDKN2A), for genetic changes by MLPA has been used to evaluate genetic heterogeneity of uveal melanomas [95] and to distinguish between Sptiz nevi and atypical spitzoid melanocytic tumors [96], as it can be difficult to distinguish these two lesion based on histology alone.

Identification of genetic alterations in melanoma tumor samples and cell lines provides investigators with pertinent information regarding genetic alterations which may contribute to melanoma pathogenesis, which also has the potential to lead to development of novel targeted therapeutics. Additionally, detection of genetic events known to be associated with melanoma can help to guide clinical decisions and treatment plans in the setting of indeterminate lesions.

Massively parallel sequencing

The development of massively parallel sequencing (MPS), also referred to as next-generation sequencing, has revolutionized the way in which DNA from tumor samples is analyzed. MPS allows for the analysis of whole genomes, exomes, or targeted regions (i.e. select genes) in individual tumor samples providing simultaneous information regarding mutational analysis of a wide range of genes and identified mutations, genetic alterations (including deletion and insertions), and copy number gains and losses. A number of different platforms are available to perform massively parallel sequencing such as the HiSeq™2000, HiSeq™2500 and miSeq (Illumina, San Diego, CA) and IonTorrent™ and IonProton™ (Life Technologies, Grand Island, NY), which are reviewed in detail by Ross and Cronin [97]. To perform MPS, DNA libraries are prepared from individual samples. In brief, genomic DNA is sheared to 150–200 bp fragments, blunt ended, and ligated with tagged adaptors and indexes (bar codes), which allow for sample identification. Optimal fragment sizes of DNA within these libraries depend on the length of the sequence reads. DNA libraries are combined with capture baits for targeted sequencing and whole exome sequencing (WES) or remain uncaptured for whole genome sequencing (WGS). The DNA then is sequenced, undergoing amplification and repetitive cycles of sequencing and detection. With improving technology, an increasing number of samples can be multiplexed while retaining mutation detection capability. The ideal read depth, which is the number of sequence reads of a particular nucleotide, varies depending upon whether whole genome, whole exome, or targeted MPS is being done [98].

Somatic mutations in tumor DNA can be challenging to identify given possible admixture of surrounding normal cells and tumor heterogeneity. Identification of low frequency mutations is crucial in the characterization of all tumor samples, including melanoma tumor samples, as it has implications for treatment options, including targeted therapies. Initial platforms for MPS were higher in cost per sample compared to traditional sequencing techniques, which was prohibitive for running large number of samples. However, over time, as technology has advanced, the cost per sample has decreased making it more attractive to use these methodologies to analyze multiple tumor DNA samples. Initially, the source DNA was restricted to fresh frozen tumor samples; however, several studies have demonstrated that adequate results can be achieved using FFPE tumor samples [98–101]. Despite these advances, DNA quality remains a crucial determinant of MPS success.

In addition to whole genome and whole exome analysis, massively parallel sequencing with targeted capture is also used to evaluate tumor samples focusing on specific genes of interest. A number of targeted capture platforms are commercially available to test for common cancer somatic mutations, including such examples as TruSeq Amplicon Cancer Panel (Illumina, San Diego, CA) and Somatic Mutation Analysis (SOMA) panel (Ambry Genetics, Aliso Viejo, CA). These targeted captures provide the advantage of deep sequencing of select, known genes. Of note, whole exome captures generally only select for ~85% of the complete exome [102–105], so if there is poor coverage over your gene of choice, it will provide limited information.

An immense amount of data is generated from massively parallel sequencing and analysis remains a challenge. The softwares available for data analysis are constantly evolving. Moreover, methodologies used for analysis also depend upon whether germline or somatic genomes are being sequenced. Mutations are first identified and then annotated in order to best assess their potential function. Briefly, sequence data is aligned to the human genome most commonly with the Burrows-Wheeler Aligner (BWA) [106, 107]. Variants are found using programs which detect single nucleotide variants (SNVs), as compared to the reference sequence, as well as insertions and deletions (indels), though programs such as the Genome Analysis Toolkit (GATK) [108] and Pindel [109]. The analysis of genomic rearrangements and copy number alterations for targeted massively parallel sequencing lags behind that of SNVs and small indels, but are evaluated using programs specific to these types of genetic aberrations, such as VarScan2 [110, 111]. Annotation with programs, such as ANNOVAR, provides information regarding the potential function of identified genetic variants [112]. ANNOVAR calls variants as frameshift indel, non-frameshift indel, stopgain, stoploss, synonymous, non-synonymous and splicing (intronic and exonic). ANNOVAR automatically identifies variants previously reported in pubic databases, including EVS6500, 1000 Genome (1000G), dbSNP (Flagged/ Nonflagged) and COSMIC [113]. ANNOVAR also annotates SNVs using SIFT, Polyphen2, MutationTaster and PhyloP to make predictions about function [113–118]. Mutation information obtained using ANNOVAR can be used to filter variants based on specific score cutoffs for the different software programs. The pipeline for mutation identification and annotation will differ depending on input DNA, that is, germline DNA versus tumor DNA. Software has been developed specifically for the analysis of somatic genomes and mutations including BreakPointer, Indelocator, and MuTect (www.broadinstitute.org/cancer/cga/). MuTect is a sequence analysis program that uses the sequence of both normal and tumor to identify somatic point mutations [119]. Despite technology and software advances, the pathogenicity of a number of the detected genetic variants, both germline and somatic, will have unknown significance. These variants of unknown significance pose challenges for clinicians as these variants are not clinically actionable and it is not clear whether these variants are involved in tumor pathogenesis.

Results of whole exome sequencing/whole genome sequencing in melanoma

Melanoma tumor samples have been evaluated using whole genome and whole exome sequencing. An initial whole genome sequencing study identified 33, 345 somatic mutations, 680 deletions, 303 insertions, and 51 rearrangements in a single melanoma cell line derived from metastatic melanoma when compared to matched germline DNA [22]. Whole genome sequencing detected known somatic mutations involved in melanoma pathogenesis including BRAF V600E, PTEN deletion, and a two base pair deletion within CDKN2A. Potential driver mutations were also identified in transcription factors, including SPDEF; genes thought to be involved in metastasis, including MMP28; and proposed tumor suppressor genes, including UVRAG [22]. Wei et al. [62] described the identification of recurrent mutations within TRRAP in 4% (6/167) metastatic tumor samples with functional studies of TRRAP suggesting it functions as an oncogene. Additional somatic mutations were identified in GRIN2A, which was mutated in 33% (17/52) of melanoma samples [62]. Somatic mutations in GRIN2A also were identified by whole genome sequencing of a melanoma tumor sample/normal DNA pair [41], but other somatic mutations suggested by the study have not been validated in subsequent massively parallel sequencing analyses. In addition to these novel genes, whole exome sequencing detected known somatic mutations including BRAF mutations in 50% of samples, consistent with previously published observations. However, no NRAS mutations were identified in these melanoma samples, in contrast to the multiple publications showing a frequency of mutations in 15–20% of melanomas [42–44]. Additional whole exome studies also identified gain of function mutations in genes found in pathways known to be involved in melanoma pathogenesis, such as MAP2K1 and MAP2K2 [61]. Evaluation of an expanded panel of melanoma samples identified mutations within these two genes in 8% (10/127) samples. Additional previously unidentified somatic mutations were observed in FAT4, DSC1, and LRP1B, but their role in melanoma pathogenesis in unknown [61]. However, it is important to note that subsequent studies have not validated the FAT4, DSC1, and LRP1B mutations in independent analysis of multiple melanoma tumor samples.

Two recent studies using whole exome sequencing of a large number of samples generated a more comprehensive understanding of the genetic landscape of somatic mutations in melanoma [20, 21]. Hodis et al. [20] reported on the results from whole exome sequencing analysis of 121 melanoma/normal DNA pairs. In this study, the authors used a statistical approach comparing the frequency of mutations in intron sequences adjacent to exon sequences to identify novel driver mutations in melanoma. Six genes demonstrated recurrent somatic mutations novel in melanoma. Activating mutations were described in PPP6C, catalytic subunit of PP6 protein phosphatase and potential tumor suppressor [20, 57]; RAC1, member of Rho family of GTPases [58, 120]; SNX31, protein sorting nexin 31, a possible Ras effector protein [121]; TACC1, transforming acidic coiled-coil protein 1 which potentially stimulates Ras and PI3K pathways [122]; and STK19, a predicted kinase, generally clustered around hotspot regions. Loss of function mutations were observed in ARID2, component of the SWI/SNF chromatin remodeling complex [123]. In addition to these novel somatic mutations, mutations were identified in known genes, such as BRAF, NRAS, PTEN, TP53, CDKN2A, and MAP2K1 [20]. All mutations were identified in over 4% of melanoma samples. Whole exome sequencing also was done by Halaban and colleagues at Yale University, in 147 melanoma samples, either primary melanomas or metastases. They also identified novel somatic mutations at higher rates in NF1, PPP6C, RAC1, and ARID2. In addition, Krauthammer et al. [21] also identified additional novel somatic mutations in melanoma samples in PTPRK, protein tyrosine phosphatase, receptor type K; PTPRD, protein tyrosine phosphatase receptor type D; and DYNC1I1, dynein, cytoplasmic 1, intermediate chain 1, which may be involved in chromosomal segregation [124]. Some of these newly identified driver mutations are associated with BRAF/NRAS mutations, but others have been specifically identified in melanomas lacking these mutations.

In addition to the individual examination of melanoma tumor samples by investigators, mutational data on a large number of melanoma tumor samples, from tumor metastases, are publically available through the Cancer Genome Atlas (TCGA) (http.//www.cbioportal.org/public-portal/; http.//gdac.broadinstitute. org/; and https.//tcga-data.nci.nih.gov/tcga/), and are continuing to be collected.

Currently, the skin cutaneous melanoma (SKCM) TCGA dataset reports the results of available mutational analysis of 337 biospecimens from metastatic melanoma tumor samples including somatic mutations, copy number, methylation clustering, protein activities, and gene expression analyses. Within the skin cutaneous melanoma (metastatic) dataset, specific copy number changes can be identified as well as somatic mutations and the types are available, highlighting those that are UVinduced. Co-mutation plots provide information regarding simultaneous mutations in different samples, allowing for grouping of melanoma tumor samples. Moreover, the available dataset can be queried to investigate specific genes of interest, either singly or for pathway analysis, and results provide information regarding copy number alterations and somatic mutations. Detailed information regarding specific types of somatic mutations are available, and provide insight into types of mutations commonly identified within a particular gene. mRNA and protein expression data is provided along with methylation profiling. The TCGA endeavor undertaken by a number of collaborators provides a large dataset of metastatic melanoma tumor samples and subsequent analysis in one central repository, making it available to all investigators to use this information for research and clinical purposes.

[contact-form-7 id=»5168″ title=»Контактная форма 1″]

0

Добавить комментарий

Войти с помощью: 

Ваш e-mail не будет опубликован. Обязательные поля помечены *