|
Review Article
Whole genome sequencing for lung cancer
1Thoracic Research Laboratory, Department of Thoracic Medicine, The Prince Charles Hospital, Brisbane, Australia; 2UQ Thoracic
Research Centre, School of Medicine, The University of Queensland, Brisbane, Australia; 3Pathology Queensland, Department of
Anatomical Pathology, The Prince Charles Hospital; Brisbane, Australia
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Abstract
Lung cancer is a leading cause of cancer related morbidity and mortality globally, and carries a dismal prognosis. Improved
understanding of the biology of cancer is required to improve patient outcomes. Next-generation sequencing (NGS) is a
powerful tool for whole genome characterisation, enabling comprehensive examination of somatic mutations that drive
oncogenesis. Most NGS methods are based on polymerase chain reaction (PCR) amplification of platform-specific DNA
fragment libraries, which are then sequenced. These techniques are well suited to high-throughput sequencing and are
able to detect the full spectrum of genomic changes present in cancer. However, they require considerable investments in
time, laboratory infrastructure, computational analysis and bioinformatic support. Next-generation sequencing has been
applied to studies of the whole genome, exome, transcriptome and epigenome, and is changing the paradigm of lung cancer
research and patient care. The results of this new technology will transform current knowledge of oncogenic pathways and
provide molecular targets of use in the diagnosis and treatment of cancer. Somatic mutations in lung cancer have already
been identified by NGS, and large scale genomic studies are underway. Personalised treatment strategies will improve care
for those likely to benefit from available therapies, while sparing others the expense and morbidity of futile intervention.
Organisational, computational and bioinformatic challenges of NGS are driving technological advances as well as raising
ethical issues relating to informed consent and data release. Differentiation between driver and passenger mutations requires
careful interpretation of sequencing data. Challenges in the interpretation of results arise from the types of specimens used
for DNA extraction, sample processing techniques and tumour content. Tumour heterogeneity can reduce power to detect
mutations implicated in oncogenesis. Next-generation sequencing will facilitate investigation of the biological and clinical
implications of such variation. These techniques can now be applied to single cells and free circulating DNA, and possibly
in the future to DNA obtained from body fluids and from subpopulations of tumour. As costs reduce, and speed and
processing accuracy increase, NGS technology will become increasingly accessible to researchers and clinicians, with the
ultimate goal of improving the care of patients with lung cancer.
Key words
High-throughput nucleotide sequencing; DNA sequence analysis; lung neoplasms; non-small cell lung carcinoma; small
cell lung carcinoma
J Thorac Dis 2012;4(2):155-163. DOI: 10.3978/j.issn.2072-1439.2012.02.01 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Introduction
Lung cancer is the most common cause of cancer death globally
(1,2), and was responsible for 1.4 million deaths in 2008 (3).
Five year survival rates remain around 15% (1,2), with the
majority of cases at an advanced stage at the time of diagnosis
(1). Improved understanding of the pathogenesis of lung cancer
is required to aid timely diagnosis, selection of cancer treatment
and development of new therapeutic modalities in order to
improve patient outcomes. The search for the genomic basis
of cancer has expanded exponentially since the introduction
of DNA sequencing techniques in the late 1970s (4,5) and identification of the first naturally occurring human cancercausing
somatic mutation in the early 1980s (6,7). Following
these discoveries, it quickly became apparent that cancer gene
discovery could either continue in a piecemeal fashion or the
whole cancer genome could be sequenced (8). In 2004, further
technological advances led to sequencing of the first, 99.7%
complete, human genome (9,10). The multi-centre, resourceintensive
Human Genome Project took many years to complete
using first-generation sequencing techniques, at an estimated
cost of $10-25 million, and generated much debate about the
efficiency of existing technology (11).
Substantial advances made since 2004 have had a major
impact on our ability to explore the genome. Traditional
techniques such as capillary electrophoresis-based DNA
sequencing, genome-wide analysis of amplifications and
deletions using array-based technology, and gene expression
arrays have been used to identify genomic drivers of lung cancer
(12-14) that can be targeted by directed therapies (12,15).
However, with second-generation or next-generation sequencing
(NGS) technology, it is now possible to sequence complete
genomes (16), exomes (17) and transcriptomes (18). These
technologies are revolutionising how we explore the genome
and exponentially increasing scope for investigation of cancer
pathogenesis, diagnosis and treatment (19-21). | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Next-generation sequencing techniques
Next-generation DNA technology has been commercially
available since 2004. These massively parallel sequencing
platforms share many advantages over traditional techniques.
Sample preparation for sequencing is streamlined, and the yield
of sequence reads is considerably greater than that possible
using capillary sequencers. Each method requires significant
investments in laboratory infrastructure and computational
support, but these massively parallel techniques provide
exponentially greater sequencing capabilities than firstgeneration
technology (19).
Many available sequencing techniques share conceptual
similarities. The Roche/454 Pyrosequencer, Illumina HiSeq and
Genome Analyser, and Applied Biosystems SOLiDTM require a
DNA fragment library, obtained by annealing platform specific
linkers to DNA fragments generated from the genome of interest.
Each strand in a fragment library is amplified by PCR prior to
performing sequencing reactions on the amplified strands (19).
The utility of NGS platforms for various applications depends
on the nature of DNA reads obtained, read time and error rates,
cost, data processing capacity and computational requirements
for analysis (see Table 1).
Driven by the potential to translate biological discovery into
improvements in patient care, NGS techniques are constantly
improving and expanding. Newly developed methods include
the non-optical Ion Torrent platform (26), and combinatorial
probe-anchor ligation sequencing, in which fluorescence-labelled
nucleotides are incorporated into sequences generated from
template DNA that has been aggregated into nanoballs (27).
Illumina HiSeq and Genome Analyser Systems
The massively parallel platform ultimately acquired by Illumina
after a sequence of company mergers was based on the second or next-generation DNA sequencing technique. After the
creation of a DNA fragment library from the target genome
sequence, both ends of each fragment are ligated to an Illumina
specific adaptor. A flow cell containing eight individual lanes
populated with capture oligonucleotide anchors hybridises the
modified DNA fragments. DNA amplification then takes place
by bridge amplification, in which the DNA template arches over
and hybridises to an adjacent oligonucleotide primer anchor
complementary to the adaptor sequence attached to the template
DNA. After amplification, sequencing commences with the
addition of DNA polymerase and a mixture of four differently
coloured fluorescent reversible dye terminators in the flow cell.
DNA fragments are extended one nucleotide at a time. Digital
images are acquired after each addition to record the location of
the labelled nucleotide. Finally, dye and terminal 3’ blockers are
removed prior to the next cycle of nucleotide coupling (19,21).
Although subject to nucleotide incorporation errors and
imperfect polymerase activity, advances in bioinformatic
analysis techniques have improved the accuracy of sequence
reads (28,29). Illumina provides the most widely used NGS
technology (21), and this has found application in whole
genome (30), and whole exome sequencing (31) including The
Cancer Genome Project (32).
454 FLX Pyrosequencer
This was the first massively parallel DNA sequencing technique to
be introduced commercially. DNA library fragments are attached
to 454-specific adaptor sequences, then mixed with agarose beads
carrying oligonucleotides complementary to the adapter sequence.
Each DNA fragment anneals to one bead, and individual fragment:
bead complexes are placed in an oil: water micelle containing PCR
reactants. Using thermal cycling, one million copies of each DNA
fragment are formed on each bead. Beads are then placed in a
picotitre plate and each remains in a fixed location for sequencing.
After addition of enzymes to catalyse the pyrosequencing
reactions, pure single nucleotide solutions are sequentially
introduced. Pyrophosphate released by DNA polymerase at each
nucleotide incorporation initiates a series of reactions ultimately
producing an amount of light proportional to the number of
nucleotides incorporated. Image acquisition after each nucleotide
incorporation step permits analysis of the DNA sequence (19,33).
Pyrosequencing has been used in the detection of single
nucleotide polymorphisms (34) and targeted capture sequencing
(35). Most inaccuracies are due to limitations of the PCR
technique used for DNA amplification, and errors in sequencing
homopolymer repeats greater than seven bases in length (33).
However, this technique is able to generate longer reads in less
time than other NGS, and improvements in experimental and
analytical techniques are providing ways to improve sequence
accuracy (33).
SOLiDTM sequencing
Supported Oligonucleotide Ligation and Detection relies on
PCR amplification of template DNA fragments on the surface of
1μm magnetic beads. The fragments are deposited onto a flow
cell slide, with two flow cells per instrument run, each able to
contain four DNA libraries. Primer is annealed to the adaptor
sequences on each amplified DNA fragment. Specific fluorescent
octamers whose fourth and fifth bases are encoded by an attached
fluorescent label are then incorporated into the DNA sequence by
DNA ligase. Di-base probes interrogate every first and second base
in each ligation reaction. The ligation is followed by fluorescence
detection, removal of the fluorescent group, then a further round
of ligation. After a series of five ligation cycles, the extension
product is removed and prepared for further ligation cycles with a
primer complementary to the n-1 position (19,21,36).
SOLiDTM technology has similar applications to the Illumina
platform. While base calling errors are reduced by two base
encoding, this is at the expense of test speed and analytical
simplicity (21,36).
Ion Torrent
This technique eliminates the complexity of the optical detection
systems of many NGS methods. DNA polymerase sequencing
with unmodified nucleotides occurs, and hydrogen ions released
during each cycle of polymerisation are detected using a
semiconductor (26).
As this technique has low capacity for parallel sequencing, it is
primarily used for sequencing short genomic segments of interest.
Although there have been issues with error rates, this is the fastest
and lowest cost NGS available (21), and sequencing capacity
using this platform is increasing exponentially. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Next-generation sequencing vs. sanger sequencing
Next-generation sequencing techniques provide many potential
advantages over traditional first-generation sequencing techniques
such as Sanger sequencing. However, Sanger sequencing remains
the method of choice for detecting genetic changes in a patient’s
genome for the purposes of guiding therapy, both in lung cancer
(37) and other malignancies (38). Sanger sequencing is used to
detect EGFR mutations in non-small cell lung cancer (NSCLC)
patients to select those who may benefit from targeted therapy
(37). Similarly, fluorescence in-situ hybridisation has been used
to detect EML4-ALK translocation in trials of Crizotinib in lung
cancer (15).
Sanger’s chain termination method of DNA sequencing (4) is
based on automatic detection of fluorescence-labelled nucleotide sequences. DNA elongation occurs along single-stranded DNA
templates and is randomly terminated by incorporation of
fluorescent dideoxynucleotide chain terminators. DNA sequences
of increasing length are detected by capillary electrophoresis.
This technique is unable to detect structural changes such as
translocations, or gene copy number changes, and multiplexing
is difficult and costly. On the other hand, NGS is well suited to
large scale gene sequencing and can provide this at lower cost per
base than traditional techniques (21). Illumina and SOLiDTM
techniques are able to yield tens of millions of reads, and the
Roche/454 platform several hundred thousand, compared with
the 96 reads produced by a capillary sequencer run (19).
While cost per base is lower for massively parallel sequencing,
the cost per test remains greater than for traditional methods.
Longer run times are required for NGS due to the need for
more frequent image acquisition than Sanger sequencing, and
reads are much shorter (21). Platform specific investments in
laboratory information management, computational analysis and
bioinformatic support are required to produce and interpret final
sequence reads, taking into account the unique error model of
each platform (19).
Next genome sequencing is necessitating a paradigm shift
in the organisation required for genomic sequencing and the
information technology and laboratory systems required to
support it. Application of these techniques to the study of not
only the whole genome, but also exomes (39), transcriptomes
(40), and epigenetics (41) will extend their scope for scientific
discovery (19). | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Next-generation sequencing in lung cancer
Prior to the introduction of NGS, candidate gene studies (12,42)
had begun to provide insight into the genomic drivers of lung
cancer, such as mutations in BRAF (42) and EGFR (37). The
introduction of massively parallel sequencing, however, has
seen further discoveries of somatic mutations in lung cancer,
with scope for an exponential increase in understanding of
oncogenesis (Table 2).
The first application of massively parallel sequencing in lung
cancer research was published in 2008. Examination of genomes
from two lung cancer cell lines characterised 306 germline
structural changes and 103 somatic rearrangements with single
base resolution, including four associated with changes in gene
expression (30). In 2010, Lee et al. published whole genome
studies of tumour and normal lung specimens obtained from
a patient with an adenocarcinoma. A large set of new somatic
mutations were discovered, at a rate of around 17.7 per megabase
of template DNA sequenced. Mutations were located in nonexpressed
genes and promoter regions up to five kilobases
upstream of coding proteins rather than within expressed genes.
Five hundred and thirty somatic single nucleotide variants were
found, including one in the KRAS proto-oncogene, and 391
in other coding regions. Forty-three structural variations were
detected (27).
Pleasance et al. published the results of massively parallel
sequencing of a small cell lung cancer cell line the same year.
Of the 22,910 somatic mutations found, 134 were detected in
coding regions. A tandem duplication of exons 3-8 of CHD7 in
frame was identified, and mutation signatures associated with
tobacco exposure were documented (43).
Most recently, examination of paired NSCLC and normal
lung tissue from a never smoking patient with adenocarcinoma
by Ju et al. led to the discovery of a novel fusion gene, KIF5BRET.
Correlation between whole genome and transcriptome
sequences from this tumour demonstrated overexpression of
chimeric RET receptor tyrosine as well as the chromosomal inversion (44).
These results, including the identification of single nucleotide
mutations as well as large structural changes, testify to the
utility of NGS in comprehensively characterising the genomic
changes present in cancers. However, most mutations present
in tumour DNA are merely passengers and do not contribute to
oncogenesis (45). Successful identification of driver mutations
amongst clusters of random passenger mutations requires the
power of large sample sizes. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The cancer genome atlas
An international effort, led by the International Cancer Genome
Consortium (ICGC), is underway to comprehensively catalogue
the genomic and epigenomic changes in cancer (46). The
ICGC is coordinating cancer genome studies in 50 different
cancer types that are of global clinical and societal importance.
A contributor to this project, The Cancer Genome Atlas
(TCGA), is a multinational, collaborative project that seeks to
apply NGS technology to further scientific knowledge of the
biology of cancer, with the aim of improving cancer care (47).
It is coordinated by a consortium between the National Cancer
Institute, the National Human Genome Research Institute and
several international specimen source centres and analytical sites
across the globe. More than 20 cancer types have been selected
for study by the TCGA to contribute to the ICGC, with up to
500 samples to be collected for each tumour type, including lung
adenocarcinoma and squamous cell carcinomas.
This large scale project requires the coordinated efforts of
several specialised TCGA research network components. To
ensure nucleotide sequencing of the highest quality, paired
tumour and normal control samples are examined by the
Biospecimen Core Resource laboratories to confirm specimens
meet stringent standards for tumour quality and quantity.
Extracted DNA and RNA are sequenced, the results analysed
and genome changes then interpreted in the context of clinical
data at Genome Characterisation, Sequencing, and Data
Coordinating Centres. Data are made publically available to the
international community to facilitate further data analysis and
validation of discoveries (47).
The Cancer Genome Project has provided impetus
for advances in experimental, computational analysis and
bioinformatic research methods to ensure efficient processing of
the large number of samples required and data obtained (48). | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Clinical applications of next-generation sequencing
The ultimate goal of elucidating the mechanisms of
cancer pathogenesis is to improve strategies for diagnosis,
prognostication and treatment of patients with cancer. Detailed
molecular subtyping at the time of diagnosis permits selection
of personalised therapies for patients who are most likely to
benefit. This benefits not only the individual receiving treatment
but spares others, and the community, the cost and morbidity
of futile intervention (49,50). Next-generation sequencing may
also be incorporated into patient selection algorithms for clinical
trials. Improved techniques for tumour categorisation in clinical
trials will increase statistical power to detect clinically important
results (50).
A thorough understanding of the molecular drivers of
cancer may ultimately allow prediction of an individual’s risk of
developing cancer on the basis of their constitutional genome
sequence. The presence of genetic variants and non-genetic
contributors to cancer phenotypes will limit the precision of
such prediction. Nevertheless, discovery of mutations and
chromosomal changes predictive of disease may permit risk
reduction and prevention strategies for some cancers (50).
The widespread application of NGS, both in research
and clinical settings, raises many practical and ethical issues.
Experiments are expensive and the results, to some extent,
unpredictable. Institutional ethical approval, patient consent,
intellectual property and data release all require careful
consideration. Given the costly and demanding nature of NGS,
coordinated applications such as TCGA are critical in order to
maximise use of resources (51).
Issues for Clinical Translation
Sample Processing
Translating the discoveries obtained by whole genome
sequencing into clinical practice will require thoughtfully
designed clinical trials that incorporate genomic studies into the
patient care algorithm. Procurement of high quality, tumourrich,
fresh frozen specimens with matched blood samples from
consented patients is critical, as is optimal specimen processing
to ensure the absence of artefacts in the sequencing data
obtained.
Specimens require storage under conditions which maintain
the integrity of nucleic acids (52).
Tissue samples are often formalin-fixed and paraffinembedded
(FFPE). While DNA extracted from such specimens
may still be suitable for whole genome sequencing, FFPE
processing, especially fixative concentration and pH, and
duration of FFPE storage may disturb DNA integrity (53,54).
The technical challenges of using archived FFPE samples for
direct DNA sequencing were highlighted during reanalysis of
the landmark BR.21 study, which used direct DNA sequencing
to screen for EGFR mutations (55). Mutational analysis was
attempted on 197 samples, of which 40 (20.3%) required
microdissection to obtain sufficient tumour for analysis. Due to technical difficulties, mutation analysis was not possible in 20
(10%) samples (55).
Whole genome sequencing must overcome challenges
posed by clinical samples, which may be of suboptimal size,
contain necrotic tumour, have high stromal or non-tumoral
content, or yield fragmented DNA. These limitations have
prompted the development of novel experimental approaches
and computational methods in order to obtain accurate secondgeneration
sequencing data from FFPE samples (56,57).
Tumour Content in Diagnostic Samples
Whole genome sequencing using NGS is typically performed on
surgically resected tumour samples (58,59). However, as most
patients are diagnosed at an advanced stage, such specimens
are obtained in only a minority of patients with NSCLC (60).
In situations where surgically resected specimens are available
for whole genome sequencing, DNA quality from fresh/frozen
samples is better, yielding more accurate sequencing data than
tissue processed by FFPE (61,62).
In patients with advanced lung cancer, only diagnostic
biopsy samples that can be obtained with minimal morbidity
are available for molecular testing. Diagnosis for these patients
is often based on small biopsy samples obtained during
bronchoscopy or percutaneous needle aspiration biopsy
procedures. Limited tumour tissue is obtained in such samples.
A recent study used computer-aided morphometry to measure
tumour area in morphologic sections of 100 bronchial biopsy
samples and found that only 48% of biopsy samples contained
tumour (63). Hence if bronchial biopsy samples are to be
used for molecular testing, they should be first stained with
haematoxylin and eosin (H&E) and the percentage of tumour
cell content estimated by a pathologist. If the tumour content
is less than 40%, the sample may require microdissection to
augment the tumour cell proportion, a technique likely to
be impractical in the clinical setting (64,65). Such additional
processing steps are constrained by the amount of sample
available.
The nature and purity of tumour in samples used as a source
of DNA for NGS has the potential to profoundly influence
interpretation of the sequencing output. Refinement of
bioinformatic techniques to assess the significance of a somatic
mutation in the context of the factors such as the background
mutation rate, tumour ploidy and stromal contamination for
primary cancers remain a major challenge of NGS (20).
Tumour Heterogeneity
Not only is the tumour content of specimens used to source
DNA of great importance, but heterogeneity within tumour
samples can also have a profound impact on NGS results (20).
Intertumoral heterogeneity has long been recognised, as
reflected in tumour classification systems based on morphology
and immunohistochemical profiles. Cognisant of intratumoral
heterogeneity, pathologists inspect several tumour sections when
thoroughly examining a specimen. Intratumoral heterogeneity is
particularly prevalent in pulmonary adenocarcinoma, leading to
the recent recommendation by an international expert panel of
quantitation of the predominant and lesser histological patterns
in pathology reports to better predict prognosis and guide
subsequent molecular investigation (66). DNA sequencing
studies have provided insight into the molecular basis of this
tumour heterogeneity. In a study of brain metastases from breast
cancer by Ding et al. (67), similar sets of coding mutations were
identified in the primary tumour and the metastasis, with gross
differences in allelic frequency. This finding was suggestive of
the presence of small subpopulations of cells with metastatic
potential in the primary tumour.
The presence of subpopulations adds further complexity to
cancer genome sequencing, and requires further investigation
(68). Intratumoral heterogeneity may influence tumour
aggressiveness, treatment responsiveness and resistance
(69). Techniques such as macro-dissection or laser capture
microdissection may permit isolation of subpopulations of cells
prior to NGS, thus increasing power to detect mutations present
in only part of the tumour (68). It is now possible to sequence
amplified DNA sequences from single cells using NGS (70).
Using this technology, it will become plausible to examine DNA
obtained in samples of body fluid such as blood, urine or pleural
fluid to diagnose cancer, and monitor for treatment resistance or
recurrence. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Summary
Next-generation technology has the potential to transform our
understanding of oncogenesis, techniques available for cancer
research and the paradigm of cancer care. Whole genome
sequencing is a powerful technique with which to examine
changes present in the cancer genome, but requires considerable
investments in sample collection and processing, computational
analysis and bioinformatic interpretation of results. Large scale,
international collaborative efforts are currently underway to
apply this technology to the study of many cancers of public
health importance.
Ongoing advances in NGS will lower the cost and increase
speed and accuracy of processing, enable its application to
smaller specimens and increase its accessibility to researchers.
Ultimately, NGS may be available to clinicians in the diagnosis,
treatment and follow-up of patients with lung cancer. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Funding sources
This work was supported by National Health and Medical
Research Council (NRMHC) project grants (KF); NHMRC Practitioner Fellowship (KF); NHMRC Career Development
Fellowship (IY); NHMRC Biomedical Scholarship (CW);
Cancer Council Queensland PhD Scholarship (MD); Cancer
Council Queensland Senior Research Fellowship (KF); Cancer
Council Queensland project grants; Queensland Smart State
project grants; Office of Health and Medical Research (OHMR)
project grants; The Prince Charles Hospital Foundation,
Australian Lung Foundation/Boehringer Ingelheim COPD
Research Fellowship (IY); NHMRC Postgraduate Medical
Scholarship (KBS); University of Queensland (UQ) PhD
Scholarship (KBS) and UQ Early Career Researcher Fellowship
(VR). | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Acknowledgements
We thank the patients and staff of The Prince Charles Hospital,
for their involvement and contribution to the TPCH lung
research program. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
References
Cite this article as: Daniels M, Goh F, Wright GM, Sriram KB, Relan V,
Clarke BE, Duhig EE, Bowman RV, Yang IA, Fong KM. Whole genome
sequencing for lung cancer. J Thorac Dis 2012;4(2):155-163. doi: 10.3978/
j.issn.2072-1439.2012.02.01
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
