Integrating next-generation sequencinginto clinical oncology: strategies,promises and pitfalls

Peter Horak, Stefan Fröhling, Hanno Glimm

To cite: Horak P, Fröhling S,Glimm H. Integrating next-generation sequencing intoclinical oncology: strategies,promises and pitfalls. ESMOOpen 2016;1:e000094.doi:10.1136/esmoopen-2016-000094

▸ Prepublication history forthis paper is available online.To view these files pleasevisit the journal online(http://dx.doi.org/10.1136/esmoopen-2016-000094).

Received 22 July 2016Revised 6 October 2016Accepted 17 October 2016

Department of TranslationalOncology, National Center forTumor Diseases Heidelberg,German Cancer ResearchCenter (DKFZ), Heidelberg,Germany

Correspondence toDr Peter Horak;peter.horak@nct-heidelberg.de

ABSTRACTWe live in an era of genomic medicine. The pastfive years brought about many significant achievementsin the field of cancer genetics, driven by rapidlyevolving technologies and plummeting costs of next-generation sequencing (NGS). The official completionof the Cancer Genome Project in 2014 led many toenvision the clinical implementation of cancer genomicdata as the next logical step in cancer therapy.Stemming from this vision, the term ‘precisiononcology’ was coined to illustrate the novelty of thisindividualised approach. The basic assumption ofprecision oncology is that molecular markers detectedby NGS will predict response to targeted therapiesindependently from tumour histology. However, alongwith a ubiquitous availability of NGS, the complexityand heterogeneity at the individual patient level had tobe acknowledged. Not only does the latter presentchallenges to clinical decision-making based onsequencing data, it is also an obstacle to the rationaldesign of clinical trials. Novel tissue-agnostic trialdesigns were quickly developed to overcome thesechallenges. Results from some of these trials haverecently demonstrated the feasibility and efficacyof this approach. On the other hand, there is anincreasing amount of whole-exome and whole-genomeNGS data which allows us to assess ever smallerdifferences between individual patients with cancer. Inthis review, we highlight different tumour sequencingstrategies currently used for precision oncology,describe their individual strengths and weaknesses,and emphasise their feasibility in different clinicalsettings. Further, we evaluate the possibility of NGSimplementation in current and future clinical trials, andpoint to the significance of NGS for translationalresearch.

FROM GENES TO GENOMES AND GENOMICSThe concept of precision oncology is notexactly a new one. Since the development ofhormonal therapies for breast cancer40 years ago,1 their efficacy was determinedbased on the predictive value of specifichormone receptor expression patterns. Thedevelopment of specific targeted therapies,including trastuzumab for breast cancer2 or

imatinib mesylate for patients with chronicmyeloid leukaemia,3 was seen as validation ofa molecularly targeted approach andinitiated the pursuit of target-driven cancerremedies. Enabled by the possibilities oflarge-scale high-throughput next-generationsequencing (NGS) technologies along withthe computational resources and tools tostore, process and analyse the data, morepowerful methods for the characterisation ofindividual patients and tumour types becameavailable. Following the first sequencedhuman cancer genome,4 NGS led to a betterunderstanding and characterisation of manycancers, resulting in definition of new sub-types, development of biomarkers and estab-lishment of novel therapeutic targets, andculminating in the completion of TheCancer Genome Atlas project (TCGA;http://cancergenome.nih.gov).5 Additionalongoing NGS endeavours includeTherapeutically Applicable Research toGenerate Effective Treatments (TARGET) forpaediatric, and International CancerGenomics Consortium (ICGC; https://dcc.icgc.org) for adult cancers. TCGA and ICGCwill generate comprehensive whole-genomesequencing (WGS) data from ∼25 000tumours. Databases such as Catalogue ofSomatic Mutations in Cancer (COSMIC)offer curated information on somatic muta-tions from more than one million tumoursamples.6 7 NGS data from these large con-sortional research efforts uncovered anumber of recurrent genomic aberrationsacross several tumour types.8 9 In addition,these studies also identified a ‘long tail’ ofrare but in many cases actionable muta-tions.10 11 Some of the knowledge acquiredfrom genome and transcriptome sequencinghas already been translated into clinicalpractice and supported the molecular sub-typing of breast cancer12 or identified noveland targetable genetic alterations in lungcancer13 14 and began influencing our

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 1

Open Access Review

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

understanding of other cancer entities.15 Higher molecu-lar resolution achieved by NGS made increasingly evidentthat most tumours harbour complex genomic changesincluding large-scale chromosomal rearrangementsdescribed depending on their origin as chromothripsis16

or as chromoplexy.17 In light of intratumoural and inter-tumoural heterogeneity, it is currently unknown howthese various events may influence the response to tar-geted treatments tailored to single driver mutations.Additionally, besides the identification of classical driverand passenger mutations,18 the occurrence of weaktumour-promoting mutations during tumour evolutionhas to be considered.19 The next important step to prac-tical application will depend on a better functional char-acterisation of the individual genetic alterations.20

Comprehensive genomic data thus empower a broadavenue of translational research projects, many of themdriven by individual research groups21 22 but also per-formed within collaborative efforts such as the CancerTarget Discovery and Development (CTD2).

THE STATE OF CLINICAL TRIALS IN PRECISION ONCOLOGYCompared to our knowledge and understanding of thecancer genome and its complexity, our current clinicalmethodologies of implementing precision oncology intoclinical practice are lagging behind. Encouraging resultsfrom non-randomised pilot studies using molecular pro-filing other than NGS to assign patients to a molecularlyguided therapy23 24 were followed by case reports on thesuccessful identification of exceptional responders byNGS in molecularly unstratified trials.25–27 The long-heldclinical suspicion that targeting an actionable and highlypredictive molecular alteration depends on the cellularcontext has recently been confirmed in patients withBRAF V600 mutations,28 having connotative implicationsfor the design of basket trials. Additional limitations of abasket trial design have been demonstrated by a phase IIstudy in thoracic tumours (CUSTOM), which accruedonly 2 out of its 15 preplanned treatment arms.29 Risingenthusiasm for precision oncology was partially curbedby the results of the French SHIVA trial, the first rando-mised, controlled, phase II study using molecular profil-ing (including NGS) and molecularly matchedtreatment in patients with advanced solid tumours.30

While this study may be pitted against a rapid integra-tion of precision medicine into clinical practice, severalcriticisms were expressed regarding its statistical design,biological rationale and clinical implications.31 First, thistrial was stratified to three therapeutic arms, with themajority of patients receiving either hormonal therapyor mTOR inhibitors, which display only limited activityas single agents outside from their specific indications.Second, the assignment to some of the targeted treat-ments was based on an inappropriate biological ration-ale. Third, two recent meta-analyses found thatmolecularly targeted therapies do indeed lead to better

outcomes when compared with non-targeted therapies.A comprehensive analysis of 570 phase II, single-agentstudies showed that those using personalised approachreported better outcomes and fewer toxicities.32 In add-ition, when considering registration trials for 58 Foodand Drug Administration (FDA)-approved anticancerdrugs from 1998 to 2013, the use of a personalised strat-egy was independently associated with higher responserates and improved progression-free and overall sur-vival.33 However, the findings of the SHIVA trial under-line the need for a better characterisation of moleculardrivers before assigning patients to targeted therapy inorder to observe a clinical benefit. This result shouldalso prompt a thorough reconsideration of the way clin-ical trials are conducted and analysed in the era of preci-sion medicine. Innovative developments includepioneering adaptive clinical trial designs based onmolecular rather than histological stratification as well asimplementing computational algorithms based on theentirety of individual patient data,34 the latter being sup-ported by comprehensive collaborative efforts such asICGCmed (http://icgcmed.org/) or CancerLinQ.35

This paradigm shift leads to a unique outlook for preci-sion medicine, in which diverse data (eg, clinical,molecular, pharmacological) are obtained and analysedin real-time, enabling rapid-learning algorithms toperform and learn from countless N-of-1 trials, andquickly extrapolate the obtained results to otherpatients. Eventually, precision oncology has to be com-bined with current effective treatment strategies by vigor-ous clinical testing and verification. Clinical trials needto master the ingrained limitations of precision oncologyin targeting rare molecular alterations across differenttumour entities. Their strategies include the develop-ment of novel matching algorithms,36 the use of rando-mised adaptive designs (I-SPY 2, BATTLE-2), targetedNGS gene panels in advanced solid tumours(NCI-MATCH, IMPACT 2) and off-target comparators(NCI-MPACT). Others focus on alterations in a specifictumour type (ALCHEMIST trials and Lung-MAP) or aspecific molecular alteration across tumour types(CREATE)37 to ensure the accrual of sufficient numberof patients. Despite these concerted efforts, larger colla-borations and data repositories will be needed toaddress less common histological or molecular subtypes,which otherwise might elude a statistical analysis. Onealternative is the retrospective identification of markersof exceptional response through NGS (eg, in NationalCancer Institute ‘Exceptional Responders’ study,NCT02243592), or the prospective accrual and analysisof an even greater number of patients with advancedcancer that have a potentially actionable genomicvariant (ASCO sponsored TAPUR trial, NCT02693535).In conclusion, the premise of precision oncology,namely to deliver the right treatment for the rightpatient in the right dose and at the right time remainsto be validated in clinical trials which have to adapt tothe changing understanding of tumour biology. We

2 Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

need to foster novel approaches to precision medicinein order to evaluate our therapeutic algorithms andgather the necessary evidence within longitudinalresearch programmes.

TARGETED PANEL SEQUENCING VERSUS WHOLE-EXOMESEQUENCING VERSUS WHOLE-GENOME SEQUENCINGNGS is the technology that makes precision oncology inits current form possible. Based on massively parallelsequencing of DNA with subsequent data processingand sequence alignment, NGS permits the simultaneousanalysis of multiple genetic aberrations, including singlenucleotide variants (SNVs), small insertions/deletions(indels) as well as copy number variants (CNVs) orcomplex genomic rearrangements. Although nowadayssequencing-by-synthesis is the predominant sequencingtechnology in use, multiple technologies and platformshave been developed and are commercially available.38

The scope of available strategies for cancer sequencingranges from targeted gene panels encompassing severalthousand base calls through whole-exome sequencing(WES) analysis of the ∼22 000 human protein-codinggenes (40–50 million bases) to WGS across all 3.3 billionbases of the human genome. NGS found its first clinicalapplication in germline testing for known monogenicand rare diseases by targeted panels39 while it was shownthat WES is ideally suited for the diagnosis of suspectednovel Mendelian diseases,40 41 where it shows a diagnos-tic accuracy of 25–30%.42 The clinical potential of NGSstarted to influence oncology as soon as 2011, when thefeasibility of integrating WGS, WES and transcriptomesequencing into oncological decision-making wasdemonstrated.43 Since then, a plethora of precisiononcology programmes and clinical trials using NGS havebeen developed and all NGS methods have been usedsuccessfully in a clinical setting. Here we describe theadvantages and limitations of the available NGS

strategies with regard to their usefulness in clinical prac-tice (table 1).

Targeted panel sequencingThe NGS panel assays allow for a rapid and reliable iden-tification of the most common and defined aberrationsfor precision oncology and range from panels of 20 tomore than 500 genes. Targeted panels rely on amplicon-based or hybridisation capture-based NGS, which showconsistent results in detecting SNVs and indels in arange of clinical applications.44 45 Albeit amplicon-basedmethods have a simpler workflow, they might not be suit-able for analysing larger panels or exomes.46 Targetedpanels offer the advantage of high depth as well as highoverall exon coverage (>99%). The depth of coveragerepresents the number of times a specific base has beensequenced and aligned to the reference genomewhereas exon coverage indicates the overall percentageof individual genes (exons) spanned by at least onesequencing read. These two variables are crucial factorsfor the consistent calling of sequence variations.47 Mosttargeted panels have an average depth of coverage of500× and more, here surpassing WES and WGS applica-tions by an order of magnitude. Insufficient exon cover-age in guanine-cytosine (GC)-rich or repetitive regions isless frequently observed in targeted panel NGS and canbe resolved by Sanger sequencing. NGS cancer panelsthus provide a rapid and cost-effective tool with highaccuracy, analytical sensitivity and specificity for thedetection of SNVs, indels and selected translocations,and can be quickly adapted in a clinical setting.48 49

Owing to the higher depth of coverage, they also offer alower threshold for uncovering intratumoural heterogen-eity and changes with low variant allele frequency, whichare inherent to many cancer types,50 albeit only withinpreselected cancer genes. Obvious limitations of targetedpanels include low sensitivity for detecting chromosomal

Table 1 Comparison of tumour DNA sequencing strategies

Targeted panels Whole exome Whole genome

Pro ▸ High depth of coverage

▸ Readily standardisable

▸ Rapid interpretation for clinical use

▸ Low costs

▸ Easy clinical implementation

▸ Detection of unknown variants

▸ Detection of CNVs

▸ Research applications

▸ Feasible in clinical routine

▸ Low price/performance ratio

▸ Comprehensive assessment of

cancer genomes

▸ Highest resolution of genomic

alterations

▸ SNVs in enhancer/promoter and

ncRNA regions

▸ Decreasing costs

▸ Subject to future studies

Contra ▸ Limited, ‘peephole’ observations

▸ Limited value for research

▸ Limited assessment of complex

aberrations

▸ Not fully comprehensive

▸ Lower CNV resolution

▸ Amplification or exon capture

necessary

▸ High bioinformatic effort

▸ Demanding clinical interpretation

▸ Time-consuming workflow

▸ Uncertain value for clinical

interpretation

▸ Most expensive

CNV, copy number variant; ncRNA, non-coding RNA; SNV, single nucleotide variant.

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 3

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

CNVs and complex genomic rearrangements. Since tar-geted panels examine genes with a known functionalrole in cancer pathogenesis, they have a higher chanceto find clinically relevant alterations, and lower probabil-ity to detect unknown or only marginally pathogenic var-iations. Targeted gene panels also reduce the complexity,duration and costs of bioinformatical and clinical inter-pretation and can therefore easily be implemented inclinical trial protocols. In fact, nearly all ongoing clinicaltrials of precision oncology rely on targeted gene panelNGS for the detection of actionable molecular targets.Most targeted panel tests can be performed on formalin-fixed paraffin-embedded (FFPE) tissue51 which reducesthe logistic and ethical burden of performing biopsies toobtain fresh tissue specimen. On the other hand, differ-ences in FFPE sample preparation and processing canhave substantial effects on the outcome of NGS. FFPE isknown for being a major source of sequence artefacts, forexample, by DNA cross-linking or cytosine deamination,52

which can be dealt with through bioinformatical or bio-chemical methods.53 Results from FFPE NGS-based panelsequencing have been shown to match up to resultsobtained from fresh frozen tissue.54 Sequencing of wholeexomes from FFPE might also be feasible.55 However, thesample age, fixation protocol and storage method seemto affect the integrity of NGS results from FFPE tissue56

and should thus be taken into account, especially wheninterpreting rare or low-frequency SNVs obtained fromold FFPE samples. Another factor that should be takeninto consideration when assessing targeted panels fromtumours without matched normal tissue is the chance ofmisinterpreting a pathogenic germline variant or report-ing a false-positive somatic alteration.57

Taken together, targeted gene panels may provide afast and cost-efficient way to obtain a somatic tumourprofile with a reasonable number of actionable altera-tions for rapid clinical interpretation. These alterationsare situated in well-defined cancer-related genes forwhich a targeted therapy is in many cases available andare therefore ideally suited for the selection and stratifi-cation of patients in clinical trials. Several targeted genepanels for somatic characterisation of solid tumours arecommercially available, and are best suited for standard-isation and validation in a routine laboratory without anysignificant bioinformatical support. They are limited bytheir omission of the vast majority of genomic informa-tion leading to non-detection of complex genomic aber-rations or mutations in genes outside of the preselectedpanel. Targeted panel sequencing is therefore restrictedto ‘peephole’ observations and offers only limitedanswers to the existing research questions.

Whole-exome sequencingWES targets approximately 1% of the whole genome. Asmost currently known disease-relevant mutations arise inprotein-coding regions, WES can detect up to 85% ofdisease-causing mutations. Using appropriate exon

capture and enrichment protocols, WES is becomingless expensive and offers a higher depth of coverage(100–150×) than WGS in up to 95% of exons. This per-centage is comparable to and more cost-effective thanWGS, leading to WES (often combined with transcrip-tome sequencing) being used as the main sequencingstrategy in some clinical settings.58 Aforementioned dif-ferences between amplicon-based and capture-basedmethods are more pronounced with regard to WES, andare reflected in their on-target rates and uniformity ofcoverage.59 In comparison to targeted panels, the clin-ical interpretation of WES is more time consuming dueto the amount of generated data. Similar constraints forpreanalytical sample preparation and preservation as totargeted panel sequencing apply.60 Nevertheless, WESfrom FFPE tissue has been shown to deliver results com-parable to fresh frozen samples in a clinically relevanttimeframe.55 In brief WES does not offer the samedepth of coverage as targeted panels and may com-pletely overlook about 5–10% of the exons.47 Yet, theoverall sensitivity of WES increases greatly with the useof paired germline testing, which can unequivocally dis-criminate between somatic and germline alterations.61

Limitations of the clinical utility of WES stem from itshigh reliance on several technological platforms introdu-cing possible biases without the possibility of rapidextrinsic validation. These range from differencesbetween exome capture methods,62–64 NGS sequencingplatforms65 to the variability of bioinformatical pipe-lines.66 With the availability of WES in a clinical setting,the identification of germline and somatic variants ofunknown significance also causes misperceptions amongphysicians as well as their patients regarding their pres-entation and interpretation.67 68 On the other hand,detection of hitherto unknown variants is one of themajor advantages of WES over targeted panel-basedsequencing as previously unrecognised cancer genesmight be discovered, prompting novel associations andhypotheses. Not only does WES provide some limitedinsights into non-exonic sequences,69 many exomecapture kits nowadays also target miRNAs, untranslatedregions and selected conserved non-coding sequences,further increasing the translational value of WES inbasic and clinical research. A compromise between tar-geted panels and WES has been suggested, consisting ofreporting only a limited panel of genes from WES forroutine clinical use, whereas additional data may beused for research and released on request to clinicians.We believe that the future of precision oncology lies inthe detailed molecular characterisation of a largenumber of patients, and a prospective linking of NGSwith individual clinical data. This, however, cannot beaccomplished by targeted panel sequencing since thelatter does not assess many sporadic genomic aberra-tions and completely fails to detect novel or unique asso-ciations. Owing to its decreasing costs, highreproducibility across experienced centres70 and a man-ageable amount of generated data suitable for rapid

4 Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

clinical interpretation, WES and transcriptome sequen-cing currently represent a ‘gold standard’ when it comesto implementing precision oncology in an academicsetting. Necessary infrastructure for clinical WES shouldbe readily available at large comprehensive cancercentres, and a clinical sequencing programme can beset up using a designated workflow (figure 1) with aturnaround time of less than 6 weeks from sample pro-cessing to clinical interpretation.

Whole-genome sequencingIn comparison to WES, WGS offers the advantage ofbeing able to uncover changes in regulatory regionssuch as promoters and enhancers, and detect intronicor intergenic rearrangements. Although WGS currentlyoffers lower depth of coverage than WES, it does notnecessitate exon capture or other enrichment strategies,thus introducing less bias in sample preparation andsequencing. At a sufficiently high coverage rate, WGSmay even outperform WES in terms of percentage ofsuccessfully sequenced exons (exon coverage).73

Although sensitivity, break-point detection and absolutecopy number estimation decline with decreasing depthof coverage, the detection of CNVs with WGS instead ofWES results in a much higher resolution.47 This againstresses the need for a consistent WGS platform, whichoffers sufficient depth of coverage (average 30–60×) overa majority of genomic regions.71 72 A perfect NGSmethod would ideally provide uniform coverage acrossthe genome without any sequence-dependent variation.However, all NGS technologies exhibit inconsistencies(low coverage or missing sequence data) in regions of

high-GC and low-GC content and long tandemrepeats.74 Owing to the sheer quantity of sequence data,this turns into a perceptible source of bias in WGS com-pared with other sequencing strategies. There are alsosignificant differences between the different WGS plat-forms regarding their genomic coverage as well as thesensitivity and comprehensiveness of the individualvariant calls.75–77 The identification of variants at lowallelic frequencies due to intratumoural heterogeneityor cancer aneuploidy requires a greater depth of cover-age,78 an obstacle which can partly be addressed bybioinformatical methods.79 Low frequency of somemolecular alterations within a sample also brings up thequestion of their driver status and their actionability,which cannot be resolved by NGS alone given the spatialand temporal limitations of current technologies.Another WGS-specific limitation stems from the preva-lence of repetitive regions and occurrence of pseudo-genes in the human genome. Short read lengths of NGSoften lead to alignment errors80 and consequently WGSmutation calling pipelines have to be methodicallybenchmarked and WGS data analysis necessitates thedevelopment of specific guidelines.72 Having said this,WGS is the platform offering the most comprehensiveand unbiased examination of the cancer genome andleads to the discovery of novel mutations81 or classifica-tions. For example, mutational signatures based on WGSand WES have been successfully used to identify possiblesubsets of patients with gastric cancer who mightrespond to PARP inhibitors.82 83 WGS allows for theidentification of driver mutations in non-coding regionssuch as untranslated regions,84 promoters,85 enhancers86

Figure 1 DKTK MASTER is an example of whole-exome and transcriptome sequencing-based precision oncology

programme. Following patient consent and study enrolment, biopsies are taken and processed in a certified laboratory using

standardised protocols and storage methods. Pathological diagnosis and tumour cell content are validated by an independent

pathologist. After DNA and RNA isolation and library preparation, NGS on an Illumina HiSeq 2500 platform is performed.

Bioinformatical analysis is followed by data curation and validation of putative molecular targets. Following discussion in a

molecular tumour board meeting, further enrolment in clinical trials and other personalised treatment strategies are

recommended. CNV, copy number variant; FISH, fluorescence in situ hybridisation; NGS, next-generation sequencing; NCT IIT,

National Center for Tumour Diseases investigator initiated trial; qRT-PCR, quantitative real-time polymerase chain reaction;

SNV, single nucleotide variant.

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 5

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

and other intergenic regulatory sequences associatedwith carcinogenesis.87 Structural variants and chromo-somal rearrangements resulting in oncogenic fusionsmay be surprisingly common in some cancer entities,88

such as prostate cancer,17 89 and can be efficientlydetected by WGS.90 91 In addition, WGS can reliablydetect genomic integration sites of pathogenic viruses,such as human papillomavirus in cervical cancer,92

which are a possible source of additional genomicinstability and copy number alterations. Apart from theidentification of variants of unknown significance in thecoding regions of the genome, interpretation of WGS isrestricted by the lack of knowledge about the functionalimpact of mutations in non-coding regions or chromo-somal aberrations.10 Furthermore, both WGS and WESmay identify mutations in splicing sites, leading to alter-native transcripts with unknown and possibly oncogenicfunctions.93 Some of the difficulties in assessing thefunctional relevance of several genomic alterations maybe resolved by integrative analysis of the transcriptome.Although it might be too early to embrace WGS in aclinical setting, this strategy provides a solid foundationfor comprehensive analysis of cancer and offers novelinsights into cancer biology leading to a better diagnosisand therapy in the future.

Limitations of NGS in precision oncologyDespite the many benefits that NGS-based methodsbrought about in the last decade, we would like to pointout the well-known as well as some less obvious generallimitations of all currently used strategies (table 1). First,there are good reasons why NGS will not replace stan-dardised and well-evidenced histopathological diagnoses.All our currently clinically applied diagnostic, prognosticand predictive tools are rooted in the histological exam-ination of the tumour. Although NGS can in many caseshelp identify and subtype different cancer entities, itshould be used in addition and not instead of an accur-ate pathological evaluation. Second, NGS from tumourbiopsies can only assess DNA and RNA changes in asmall subset of tumour cells at a given timepoint, thusproviding low temoral and spatial resolution of thewhole tumour. This problem can be tackled from manydifferent angles, such as by achieving a better spatialresolution through novel techniques,94 single-cellsequencing, serial analysis of circulating cell-free nucleicacids or tumour cells,95 96 or by pragmatically focusingon actionability of individual targets via functionalstudies. Ex vivo functional testing or assessment of circu-lating cell-free DNA97–99 might also offer some solutionsas to which alterations might be crucial for tumourgrowth or make a particularly good drug target. It isnotable, though, that many of the solutions addressingthe limitations of NGS are themselves based on NGS. Inaddition to the DNA and RNA alterations accessible byNGS, most cancers often engage in post-translational,epigenetic and metabolic adaptations to facilitate theirgrowth and metastasis. Addressing the epigenome,

proteome and metabolome may thus become the nextgreat challenge of precision oncology. Third, develop-ment of necessary software tools for analysis and clinicalinterpretation of ‘big data’ generated by NGS to supportclinical decision-making is still lagging behind the exist-ing hardware equipment for their computation, manage-ment and storage. Moreover, large bioinformatical effortis necessary in order to make data obtained on diverseNGS platforms and analysed by different bioinformaticalpipelines and algorithms directly comparable. Success ofNGS and precision oncology hinges therefore largely oneffective communication and professional collaborationbetween all parties involved.

TRANSCRIPTOME SEQUENCINGProfiling signatures based on mRNA expression are beingrapidly introduced into clinical cancer management.Beyond prognostic molecular classification,100 recentresults of the prospective TAILORx101 and MINDACT102

studies successfully validated the premise of mRNAexpression profiles as useful clinical tools for predictivetherapeutic stratification of patients with breast cancer.These multigene profiling assays use qRT-PCR-based ormicroarray-based platforms to assess mRNA expression.With regard to the genomic assessment of cancer, meas-uring gene expression in addition to DNA alterationsmight prove biologically significant and clinically useful.In the context of clinical cancer genome sequencing,RNA sequencing should be considered and performed inparallel to WES or WGS. Transcriptome sequencing pro-vides synergistic information regarding allele-specificexpression, information on transcribed gene fusions andexpression levels of cancer-specific genes.103 BesidesmRNA, several non-coding RNA (ncRNA) species includ-ing miRNA, siRNA, piRNA and lncRNA can be detectedby RNA sequencing, which increases the chances ofRNA-based biomarker discovery and development. Thedownsides of transcriptome sequencing include itsrequirement for fresh tissue along with high variability ofRNA expression levels depending on intrinsic factorssuch as cell type, cell cycle phase or cell viability. Thequantification of RNA is also susceptible to biases stem-ming from low quantities of cancer cells, heterogeneitywithin the tumour sample, and inadequate or unattain-able baseline expression reference. Transcriptomesequencing adds another layer of data, increases theburden of clinical interpretation, and is not easily stand-ardisable in clinical laboratories. To circumvent some ofthese difficulties, targeted panel strategies for RNAsequencing have been developed.104 Nevertheless, tran-scriptome sequencing provides a deeper insight into thecomplex cancer biology and will be followed by epigen-ome, proteome and metabolome assessment.

Resolving tumour heterogeneityEven the most sensitive and specific high-throughputassays deliver only a snapshot of the given specimen at a

6 Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

specific point in the evolution of cancer and patientshistory.105 Integrative analyses of large cohorts providegeneral information about mutation rates and tumourheterogeneity and while offering into clonal eventsdriving cancer progression. Nevertheless, the underlyinggenetic background and the selective pressure ofpatients’ individual medical histories and treatmentsinvariably lead to much wider heterogeneity than we areable to assess with our current methods, and might bethe main cause of treatment failure due to evolutionaryadaptation.106 107 As our sequencing data are based onlimited tumour sampling, we should not forget the factthat we are assessing only a small fraction of the individ-ual tumour. Moreover, genomic heterogeneity increasesin metastatic disease, and may further evolve by mono-clonal and polyclonal seeding as well as metastasis-to-metastasis spread.108 Owing to the plethora of possibleorigins for genomic heterogeneity and its far-reachingconsequences,109 it seems likely that precision oncologywill have to consider and cope with intratumoural het-erogeneity as one of its major limitations. There aresome reasons for cautious optimism, though.Mathematical modelling predicts that most cancersharbour 5–8 driver mutations, thus limiting the numberof possible (and druggable) culprits.8 In addition, theremight be a limited set of highly essential genes requiredfor human cancer cell proliferation110 111 although spe-cific differences between different tumour types mayexist.111 The problem of intratumoural heterogeneitymay be addressed by various methods. An unbiasedsystems biology approach may be used to rapidly identifyand target possible synergistic and vital cellular pathways.Meticulous dissection of signalling pathways in a differ-ing cellular background will be necessary to elucidateunderlying pathogenic mechanisms. Functional genom-ics may thus help us to understand the multifacetedlandscape of tumour-specific alterations with the aim ofcomprehensively characterising and understanding theheterogeneity of individual tumours. The relentlesslyinnovative pursuit of answers to research questionsraised by NGS-generated ‘big data’ will not only driveour current and future cancer research efforts, it willalso, at some point, influence our daily clinicaldecisions.

PERSONALISED NGS-BASED CLINICAL TRIAL DESIGNSThe best clinical evidence in current medical practicewas obtained through meticulously designed and thor-oughly performed randomised, controlled and prefer-ably blinded phase III clinical trials. In clinical oncology,even statistically well-performed phase III trials have tocope with the schism between statistically significant andclinically meaningful results. With the ascent of preci-sion medicine and the acknowledgment of increasingbiological complexity, many current protocol designs aredestined to become inadequate. Although many earlyclinical advances in precision oncology have been made

in entities that are considered to be molecularly homo-geneous, it is highly unrealistic to expect most solidtumours to present a single actionable mutation andnicely segregate into subtypes and thus treatmentarms.112 Targeted therapies provide a large number oftherapeutic options but in many cases, only a subset ofpatients will actually respond to targeted treatments,thus trading known common drug toxicities for a verysmall benefit. Characterising response markers retro-spectively and identifying exceptional responders is onepossible strategy of precision oncology.113 However, itdoes not provide a prospective validation necessary forhigh-level clinical evidence. Early on, prospective clinicaltrials using either ‘umbrella’ or ‘basket’ designs weredeveloped to study molecularly targeted therapies.114

The strategy of umbrella trials is to screen for a prese-lected number of molecular aberrations within a largecohort of patients who are then assigned to a targetedtreatment in a follow-up study. Histology-agnostic baskettrials focus on specific molecular alterations and enrolpatients with same molecular targets to receive the cor-responding drug. Successful clinical implementation ofbasket and umbrella trials as well their limitations havebeen illustrated recently115 and will present challengesto regulatory authorities responsible for drugapproval.116 Novel study protocols based on Bayesian sta-tistics were thus developed and include adaptive designswhich enable an ongoing modification of the clinicaltrial based on knowledge acquired through the trialitself.117 Such efforts are currently being pursued atmany institutions worldwide and may be exemplified bythe Continuous ReAssessment with Flexible exTension(CRAFT) trial design of the German CancerConsortium (DKTK). Given the increasing amount ofgenomic data and resulting combinatorial complexity,we have to venture beyond the conventional basket andumbrella approaches.118 The complexity of personalisedcancer therapy is rooted both in the large number ofpossibly druggable molecular aberrations due to intratu-moural and intertumoural heterogeneity, which resultsin increasingly oversized screening cohorts, and in theneed for rational drug combination and sequencing.However, given the increasing amounts of targetedagents, the number of potential drug combination andconceivable sequential treatments increases exponen-tially, thus making any a priori stratification for conven-tional clinical trial design impossible.A step towards truly personalised clinical trials might

lie in the implementation of oncological N-of-1 trials,119

but this approach encounters scepticism and may notprovide high-level evidence in this setting.120 Nationaland international collaborative projects which plan tointegrate genomic data with individual patients’ healthrecords, such as ASCO’s CancerLinq, AACR’s ProjectGENIE or ICGCmed, all follow a bold assumptionwhich, in theory, should equal an amalgamation of hun-dreds of thousands N-of-1 trials. Nevertheless, incorpor-ation of genomic, epigenomic, transcriptomic and

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 7

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

proteomic characterisation of tumours will invariablylead to a level of intricacy inaccessible to establishedtrial designs. A simplification of the occurring molecularconundrum might be achieved, for example, by analysingcombinations of somatic aberrations and identifyinglarger networks of key molecular changes spanning dif-ferent genes, pathways and tumour types.121 122 Rationalinclusion of immunotherapy and drug combinationsbased on actionable network nodes into clinical trialdesigns may address the insufficiency of current mono-therapies in achieving durable responses. Finally, a betterfunctional knowledge of the underlying tumour biologywill possibly enable us to devise novel, rational and adap-tive therapeutic tactics to outmanoeuvre the ever-evolvingand therapy-evading cancer in future clinical trials.

THE FUTURE OF PRECISION ONCOLOGYAs the cost of sequencing is predestined to fall evenfurther, sequencing of cancer genomes in a clinicalsetting will become more common. NGS-enabled preci-sion oncology will add another layer of complexity to ourdaily clinical decision-making. Looking at the diversity ofcancer genomes and their phenotypes, we will have togenerate high-quality NGS data and integrate them withhistopathological and clinical findings, followed by appro-priate clinical trials. Faced with the implications of thedata volume exceeding the human cognitive capacity forprocessing and interpretation in a meaningful and timelymanner, the need for rapid-learning computer-basedsystems in clinical oncology should be acknowledged.34 123

The implementation of computer-based self-learningalgorithms, such as IBM’s Watson Oncology, is currentlybeing tested at some institutions.124 125 At the other endstands a competent healthcare provider whose role is toimplement and carry out the treatment decision as wellas face its results together with the patient. While themajority of physicians in large comprehensive cancercentres regularly interpret NGS-based panels and discussthem with their patients, a quarter of them have low con-fidence in their genomic knowledge.126 One possibility toevade some challenges of a comprehensive WGS andWES assessment is to select and filter only a predefinedset of genomic data to the clinician and the patient.127

The value of reporting genetic variants with limited clin-ical significance has to be routinely discussed while guide-lines for interpretation and reporting of incidentalfindings in clinical exome and genome sequencing arebeing developed.68 128

About 150 years ago, the lifework of GregorMendel,129 titled ‘Experiments on Plant Hybridization’,was published. In this work, Mendel postulated and vali-dated many rules of heredity which enabled the develop-ment of genetics as a scientific discipline. After beingignored by the scientific community of the time, Mendelallegedly proclaimed: ‘My time will come’. It was not thecase until after his death when other scientists began to

incorporate his findings into genetics and subsequentlyapply them to other fields such as medicine. With thedevelopment of NGS and the rapid availability of indi-vidual genomic information, implementation and appli-cation of genetic data should not be slowed down byignorance and dismissal. Fortunately, many cancercentres around the world have started trials and pro-grammes of precision oncology. We should embrace thisnovel paradigm as it will most probably shape andenhance our understanding of tumour biology andcancer therapy in the decades to come. There is anurgent need to evaluate and prospectively validate ourmost promising approaches to precision oncology. Thiseffort relies on a continued participation and collabor-ation of clinical oncologists, cancer researchers, compu-tational biologists, bioinformaticians and, mostimportantly, patients. Seamless clinical implementationof precision oncology needs support from regulatoryand ethical authorities, and can only be successful withsufficient resources and long-term commitment fromnational and international funding agencies.

Contributors PH designed, wrote and reviewed the manuscript. SF and HGdesigned, commented and reviewed the manuscript.

Competing interests None declared.

Provenance and peer review Commissioned; externally peer reviewed.

Open Access This is an Open Access article distributed in accordance withthe Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license,which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, providedthe original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/

REFERENCES1. Cole MP, Jones CT, Todd ID. A new anti-oestrogenic agent in late

breast cancer. An early clinical appraisal of ICI46474. Br J Cancer1971;25:270–5.

2. Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapyplus a monoclonal antibody against HER2 for metastaticbreast cancer that overexpresses HER2. N Engl J Med2001;344:783–92.

3. Druker BJ, Talpaz M, Resta DJ, et al. Efficacy and safety of aspecific inhibitor of the BCR-ABL tyrosine kinase in chronic myeloidleukemia. N Engl J Med 2001;344:1031–7.

4. Pleasance ED, Cheetham RK, Stephens PJ, et al. Acomprehensive catalogue of somatic mutations from a humancancer genome. Nature 2010;463:191–6.

5. Weinstein JN, Collisson EA, Mills GB, et al. The Cancer GenomeAtlas Pan-Cancer analysis project. Nat Genet 2013;45:1113–20.

6. Futreal PA, Coin L, Marshall M, et al. A census of human cancergenes. Nat Rev Cancer 2004;4:177–83.

7. Forbes SA, Beare D, Gunasekaran P, et al. COSMIC: exploring theworld’s knowledge of somatic mutations in human cancer. NucleicAcids Res 2015;43:D805–811.

8. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature2009;458:719–24.

9. Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancergenome landscapes. Science 2013;339:1546–58.

10. Garraway LA, Lander ES. Lessons from the cancer genome. Cell2013;153:17–37.

11. Lawrence MS, Stojanov P, Mermel CH, et al. Discovery andsaturation analysis of cancer genes across 21 tumour types. Nature2014;505:495–501.

12. Ellis MJ, Perou CM. The genomic landscape of breast cancer as atherapeutic roadmap. Cancer Discov 2013;3:27–34.

8 Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

13. Clinical Lung Cancer Genome P, Network Genomic M. Agenomics-based classification of human lung tumors. Sci TranslMed 2013;5:209ra153.

14. Wu K, Huang RS, House L, et al. Next-generation sequencing forlung cancer. Future Oncol 2013;9:1323–36.

15. Lianos GD, Mangano A, Cho WC, et al. From standard to newgenome-based therapy of gastric cancer. Expert Rev GastroenterolHepatol 2015;9:1023–6.

16. Stephens PJ, Greenman CD, Fu B, et al. Massive genomicrearrangement acquired in a single catastrophic event duringcancer development. Cell 2011;144:27–40.

17. Baca SC, Prandi D, Lawrence MS, et al. Punctuated evolution ofprostate cancer genomes. Cell 2013;153:666–77.

18. Kandoth C, McLellan MD, Vandin F, et al. Mutational landscapeand significance across 12 major cancer types. Nature2013;502:333–9.

19. Castro-Giner F, Ratcliffe P, Tomlinson I. The mini-driver model ofpolygenic cancer evolution. Nat Rev Cancer 2015;15:680–5.

20. Ledford H. End of cancer-genome project prompts rethink. Nature2015;517:128–9.

21. Chudasama P, Renner M, Straub M, et al. Targeting FGFR1 fortreatment of soft-tissue sarcoma. Clin Cancer Res PublishedOnline First: 17 Aug 2016. doi:10.1158/1078-0432.CCR-16-0860

22. Kordes M, Roring M, Heining C, et al. Cooperation of BRAF(F595L)and mutant HRAS in histiocytic sarcoma provides new insights intooncogenic BRAF signaling. Leukemia 2016;30:937–46.

23. Von Hoff DD, Stephenson JJ Jr, Rosen P, et al. Pilot study usingmolecular profiling of patients’ tumors to find potential targets andselect treatments for their refractory cancers. J Clin Oncol2010;28:4877–83.

24. Tsimberidou AM, Iskander NG, Hong DS, et al. Personalizedmedicine in a phase I clinical trials program: the MD AndersonCancer Center initiative. Clin Cancer Res 2012;18:6373–83.

25. Iyer G, Hanrahan AJ, Milowsky MI, et al. Genome sequencingidentifies a basis for everolimus sensitivity. Science 2012;338:221.

26. Czink E, Heining C, Weber TF, et al. [Durable remission under dualHER2 blockade with Trastuzumab and Pertuzumab in a patient withmetastatic gallbladder cancer]. Z Gastroenterol 2016;54:426–30.

27. Gröschel S, Bommer M, Hutter B, et al. Integration of genomicsand histology reveals diagnosis and enables effective therapy ofrefractory cancer of unknown primary with PDL1 amplification. ColdSpring Harb Mol Case Stud Published Online First: 24 Aug 2016.doi: 10.1101/mcs.a001180.

28. Hyman DM, Puzanov I, Subbiah V, et al. Vemurafenib in multiplenonmelanoma cancers with BRAF V600 Mutations. N Engl J Med2015;373:726–36.

29. Lopez-Chavez A, Thomas A, Rajan A, et al. Molecular profiling andtargeted therapy for advanced thoracic malignancies: abiomarker-derived, multiarm, multihistology phase II basket trial.J Clin Oncol 2015;33:1000–7.

30. Le Tourneau C, Delord JP, Goncalves A, et al. Molecularly targetedtherapy based on tumour molecular profiling versus conventionaltherapy for advanced cancer (SHIVA): a multicentre, open-label,proof-of-concept, randomised, controlled phase 2 trial. LancetOncol 2015;16:1324–34.

31. Tsimberidou AM, Kurzrock R. Precision medicine: lessons learnedfrom the SHIVA trial. Lancet Oncol 2015;16:e579–580.

32. Schwaederle M, Zhao M, Lee JJ, et al. Impact of precisionmedicine in diverse cancers: a meta-analysis of phase II clinicaltrials. J Clin Oncol 2015;33:3817–25.

33. Jardim DL, Schwaederle M, Wei C, et al. Impact of abiomarker-based strategy on oncology drug development: ameta-analysis of clinical trials leading to FDA approval. J NatlCancer Inst 2015;107:djv253.

34. Shrager J, Tenenbaum JM. Rapid learning for precision oncology.Nat Rev Clin Oncol 2014;11:109–18.

35. Shah A, Stewart AK, Kolacevski A, et al. Building a rapid learninghealth care system for oncology: why CancerLinQ collectsidentifiable health information to achieve its vision. J Clin Oncol2016;34:756–63.

36. Schwaederle M, Parker BA, Schwab RB, et al. Precision oncology:the UC San Diego Moores Cancer Center PREDICT Experience.Mol Cancer Ther 2016;15:743–52.

37. Biankin AV, Piantadosi S, Hollingsworth SJ. Patient-centric trials fortherapeutic development in precision oncology. Nature2015;526:361–70.

38. Mardis ER. Next-generation sequencing platforms. Annu Rev AnalChem (Palo Alto Calif ) 2013;6:287–303.

39. Xue Y, Ankala A, Wilcox WR, et al. Solving the moleculardiagnostic testing conundrum for Mendelian disorders in the era of

next-generation sequencing: single-gene, gene panel, or exome/genome sequencing. Genet Med 2015;17:444–51.

40. Ng SB, Buckingham KJ, Lee C, et al. Exome sequencing identifiesthe cause of a Mendelian disorder. Nat Genet 2010;42:30–5.

41. Bamshad MJ, Ng SB, Bigham AW, et al. Exome sequencing as atool for Mendelian disease gene discovery. Nat Rev Genet2011;12:745–55.

42. O’Donnell-Luria AH, Miller DT. A clinician’s perspective on clinicalexome sequencing. Hum Genet 2016;135:643–54.

43. Roychowdhury S, Iyer MK, Robinson DR, et al. Personalizedoncology through integrative high-throughput sequencing: a pilotstudy. Sci Transl Med 2011;3:111ra121.

44. Kluk MJ, Lindsley RC, Aster JC, et al. Validation andimplementation of a custom next-generation sequencing clinicalassay for hematologic malignancies. J Mol Diagn 2016;18:507–15.

45. Froyen G, Broekmans A, Hillen F, et al. Validation and applicationof a custom-designed targeted next-generation sequencing panelfor the diagnostic mutational profiling of solid tumors. PLoS ONE2016;11:e0154038.

46. Hagemann IS, Cottrell CE, Lockwood CM. Design of targeted,capture-based, next generation sequencing tests for precisioncancer therapy. Cancer Genet 2013;206:420–31.

47. Sims D, Sudbery I, Ilott NE, et al. Sequencing depth and coverage:key considerations in genomic analyses. Nat Rev Genet2014;15:121–32.

48. Rehm HL, Bale SJ, Bayrak-Toydemir P, et al. ACMG clinicallaboratory standards for next-generation sequencing. Genet Med2013;15:733–47.

49. Zheng Y, Qing T, Song Y, et al. Standardization efforts enablingnext-generation sequencing and microarray based biomarkers forprecision medicine. Biomark Med 2015;9:1265–72.

50. Burrell RA, McGranahan N, Bartek J, et al. The causes andconsequences of genetic heterogeneity in cancer evolution. Nature2013;501:338–45.

51. Wagle N, Berger MF, Davis MJ, et al. High-throughput detection ofactionable genomic alterations in clinical tumor samples bytargeted, massively parallel sequencing. Cancer Discov2012;2:82–93.

52. Chen G, Mosier S, Gocke CD, et al. Cytosine deamination is amajor cause of baseline noise in next-generation sequencing. MolDiagn Ther 2014;18:587–93.

53. Do H, Dobrovic A. Dramatic reduction of sequence artefacts fromDNA isolated from formalin-fixed cancer biopsies by treatment withuracil- DNA glycosylase. Oncotarget 2012;3:546–58.

54. Spencer DH, Sehn JK, Abel HJ, et al. Comparison of clinicaltargeted next-generation sequence data from formalin-fixed andfresh-frozen tissue specimens. J Mol Diagn 2013;15:623–33.

55. Van Allen EM, Wagle N, Stojanov P, et al. Whole-exomesequencing and clinical interpretation of formalin-fixed,paraffin-embedded tumor samples to guide precision cancermedicine. Nat Med 2014;20:682–8.

56. Munchel S, Hoang Y, Zhao Y, et al. Targeted or whole genomesequencing of formalin fixed tissue samples: potential applicationsin cancer genomics. Oncotarget 2015;6:25943–61.

57. Jones S, Anagnostou V, Lytle K, et al. Personalized genomicanalyses for cancer mutation discovery and interpretation. SciTransl Med 2015;7:283ra53.

58. Roychowdhury S, Chinnaiyan AM. Translating cancer genomesand transcriptomes for precision oncology. CA Cancer J Clin2016;66:75–88.

59. Samorodnitsky E, Jewell BM, Hagopian R, et al. Evaluation ofhybridization capture versus amplicon-based methods forwhole-exome sequencing. Hum Mutat 2015;36:903–14.

60. De Paoli-Iseppi R, Johansson PA, Menzies AM, et al. Comparisonof whole-exome sequencing of matched fresh and formalin fixedparaffin embedded melanoma tumours: implications for clinicaldecision making. Pathology 2016;48:261–6.

61. Van Allen EM, Wagle N, Levy MA. Clinical analysis andinterpretation of cancer genome data. J Clin Oncol2013;31:1825–33.

62. Chilamakuri CS, Lorenz S, Madoui MA, et al. Performancecomparison of four exome capture systems for deep sequencing.BMC Genomics 2014;15:449.

63. Asan, Xu Y, Jiang H, et al. Comprehensive comparison of threecommercial human whole-exome capture platforms. Genome Biol2011;12:R95.

64. Sulonen AM, Ellonen P, Almusa H, et al. Comparison ofsolution-based exome capture methods for next generationsequencing. Genome Biol 2011;12:R94.

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 9

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

65. Clark MJ, Chen R, Lam HY, et al. Performance comparison ofexome DNA sequencing technologies. Nat Biotechnol2011;29:908–14.

66. Lelieveld SH, Veltman JA, Gilissen C. Novel bioinformaticdevelopments for exome sequencing. Hum Genet2016;135:603–14.

67. Gray SW, Park ER, Najita J, et al. Oncologists’ and cancer patients’views on whole-exome sequencing and incidental findings: resultsfrom the CanSeq study. Genet Med 2016;18:1011–19.

68. Green RC, Berg JS, Grody WW, et al. ACMG recommendations forreporting of incidental findings in clinical exome and genomesequencing. Genet Med 2013;15:565–74.

69. Guo Y, Long J, He J, et al. Exome sequencing generates highquality data in non-target regions. BMC Genomics 2012;13:194.

70. Allen EM, Robinson D, Morrissey C, et al. A comparativeassessment of clinical whole exome and transcriptome profilingacross sequencing centers: implications for precision cancermedicine. Oncotarget Published Online First: 5 May 2016. doi:10.18632/oncotarget.9184.

71. Bentley DR, Balasubramanian S, Swerdlow HP, et al. Accuratewhole human genome sequencing using reversible terminatorchemistry. Nature 2008;456:53–9.

72. Alioto TS, Buchhalter I, Derdak S, et al. A comprehensiveassessment of somatic mutation detection in cancer usingwhole-genome sequencing. Nat Commun 2015;6:10001.

73. Lelieveld SH, Spielmann M, Mundlos S, et al. Comparison ofexome and genome sequencing technologies for the completecapture of protein-coding regions. Hum Mutat 2015;36:815–22.

74. Ross MG, Russ C, Costello M, et al. Characterizing and measuringbias in sequence data. Genome Biol 2013;14:R51.

75. Lam HY, Clark MJ, Chen R, et al. Performance comparison ofwhole-genome sequencing platforms. Nat Biotechnol2012;30:78–82.

76. Rieber N, Zapatka M, Lasitschka B, et al. Coverage bias andsensitivity of variant calling for four whole-genome sequencingtechnologies. PLoS ONE 2013;8:e66621.

77. Suzuki S, Ono N, Furusawa C, et al. Comparison of sequencereads obtained from three next-generation sequencing platforms.PLoS ONE 2011;6:e19534.

78. Lee W, Jiang Z, Liu J, et al. The mutation spectrum revealed bypaired genome sequences from a lung cancer patient. Nature2010;465:473–7.

79. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive detection ofsomatic point mutations in impure and heterogeneous cancersamples. Nat Biotechnol 2013;31:213–19.

80. Treangen TJ, Salzberg SL. Repetitive DNA and next-generationsequencing: computational challenges and solutions. Nat RevGenet 2012;13:36–46.

81. Ley TJ, Mardis ER, Ding L, et al. DNA sequencing of acytogenetically normal acute myeloid leukaemia genome. Nature2008;456:66–72.

82. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures ofmutational processes in human cancer. Nature 2013;500:415–21.http://dx.doi.org/10.1038/nature12477

83. Alexandrov LB, Nik-Zainal S, Siu HC, et al. A mutational signaturein gastric cancer suggests therapeutic strategies. Nat Commun2015;6:8683.

84. Supek F, Miñana B, Valcarcel J, et al. Synonymous mutationsfrequently act as driver mutations in human cancers. Cell2014;156:1324–35.

85. Huang FW, Hodis E, Xu MJ, et al. Highly recurrent TERT promotermutations in human melanoma. Science 2013;339:957–9.

86. Mansour MR, Abraham BJ, Anders L, et al. Oncogene regulation.An oncogenic super-enhancer formed through somatic mutation ofa noncoding intergenic element. Science 2014;346:1373–7.

87. Freedman ML, Monteiro AN, Gayther SA, et al. Principles for thepost-GWAS functional characterization of cancer risk loci. NatGenet 2011;43:513–18.

88. Willis NA, Rass E, Scully R. Deciphering the code of the cancergenome: mechanisms of chromosome rearrangement. TrendsCancer 2015;1:217–30.

89. Tomlins SA, Rhodes DR, Perner S, et al. Recurrent fusion ofTMPRSS2 and ETS transcription factor genes in prostate cancer.Science 2005;310:644–8.

90. Campbell PJ, Stephens PJ, Pleasance ED, et al. Identification ofsomatically acquired rearrangements in cancer using genome-widemassively parallel paired-end sequencing. Nat Genet2008;40:722–9.

91. Yang L, Luquette LJ, Gehlenborg N, et al. Diverse mechanisms ofsomatic structural variations in human cancer genomes. Cell2013;153:919–29.

92. Ojesina AI, Lichtenstein L, Freeman SS, et al. Landscape ofgenomic alterations in cervical carcinomas. Nature2014;506:371–5.

93. Pajares MJ, Ezponda T, Catena R, et al. Alternative splicing: anemerging topic in molecular and clinical oncology. Lancet Oncol2007;8:349–57.

94. Stahl PL, Salmen F, Vickovic S, et al. Visualization and analysis ofgene expression in tissue sections by spatial transcriptomics.Science 2016;353:78–82.

95. Parsons HA, Beaver JA, Park BH. Circulating plasma tumor DNA.Adv Exp Med Biol 2016;882:259–76.

96. Meador CB, Lovly CM. Liquid biopsies reveal the dynamicnature of resistance mechanisms in solid tumors. Nat Med2015;21:663–5.

97. Friedman AA, Letai A, Fisher DE, et al. Precision medicine forcancer with next-generation functional diagnostics. Nat Rev Cancer2015;15:747–56.

98. Chi KR. The tumour trail left in blood. Nature 2016;532:269–71.99. Lianos GD, Mangano A, Cho WC, et al. Circulating tumor DNA:

new horizons for improving cancer treatment. Future Oncol2015;11:545–8.

100. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of humanbreast tumours. Nature 2000;406:747–52.

101. Sparano JA, Gray RJ, Makower DF, et al. Prospective validation ofa 21-gene expression assay in breast cancer. N Engl J Med2015;373:2005–14.

102. Cardoso F, van't Veer LJ, Bogaerts J, et al. 70-Gene Signature asan Aid to Treatment Decisions in Early-Stage Breast Cancer.N Engl J Med 2016;375:717–29.

103. Byron SA, Van Keuren-Jensen KR, Engelthaler DM, et al.Translating RNA sequencing into clinical diagnostics: opportunitiesand challenges. Nat Rev Genet 2016;17:257–71.

104. Zheng Z, Liebers M, Zhelyazkova B, et al. Anchored multiplex PCRfor targeted next-generation sequencing. Nat Med2014;20:1479–84.

105. Yates LR, Campbell PJ. Evolution of the cancer genome. Nat RevGenet 2012;13:795–806.

106. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneityand branched evolution revealed by multiregion sequencing.N Engl J Med 2012;366:883–92.

107. Mengelbier LH, Karlsson J, Lindgren D, et al. Intratumoral genomediversity parallels progression and predicts outcome in pediatriccancer. Nat Commun 2015;6:6125.

108. Gundem G, Van Loo P, Kremeyer B, et al. The evolutionaryhistory of lethal metastatic prostate cancer. Nature2015;520:353–7.

109. Marusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity:a looking glass for cancer? Nat Rev Cancer 2012;12:323–34.

110. Blomen VA, Májek P, Jae LT, et al. Gene essentiality and syntheticlethality in haploid human cells. Science 2015;350:1092–6.

111. Wang T, Birsoy K, Hughes NW, et al. Identification andcharacterization of essential genes in the human genome. Science2015;350:1096–101.

112. Bedard PL, Hansen AR, Ratain MJ, et al. Tumour heterogeneity inthe clinic. Nature 2013;501:355–64.

113. Chang DK, Grimmond SM, Evans TR, et al. Mining the genomes ofexceptional responders. Nat Rev Cancer 2014;14:291–2.

114. Sleijfer S, Bogaerts J, Siu LL. Designing transformativeclinical trials in the cancer genome era. J Clin Oncol2013;31:1834–41.

115. Baselga J. Bringing precision medicine to the clinic: from genomicprofiling to the power of clinical observation. Ann Oncol2013;24:1956–7.

116. Willyard C. ‘Basket studies’ will hold intricate data for cancer drugapprovals. Nat Med 2013;19:655.

117. Berry DA. Adaptive clinical trials in oncology. Nat Rev Clin Oncol2012;9:199–207.

118. Klauschen F, Andreeff M, Keilholz U, et al. The combinatorialcomplexity of cancer precision medicine. Oncoscience2014;1:504–9.

119. Schork NJ. Personalized medicine: time for one-person trials.Nature 2015;520:609–11.

120. Collette L, Tombal B. N-of-1 trials in oncology. Lancet Oncol2015;16:885–6.

121. Creixell P, Reimand J, Haider S, et al. Pathway and networkanalysis of cancer genomes. Nat Methods 2015;12:615–21.

122. Leiserson MD, Vandin F, Wu HT, et al. Pan-cancer networkanalysis identifies combinations of rare somatic mutations acrosspathways and protein complexes. Nat Genet 2015;47:106–14.

10 Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

123. Abernethy AP, Etheredge LM, Ganz PA, et al. Rapid-learningsystem for cancer care. J Clin Oncol 2010;28:4268–74.

124. Zauderer MG, Gucalp A, Epstein AS, et al. Piloting IBM WatsonOncology within Memorial Sloan Kettering’s regional network. In2014 ASCO Annual Meeting, J Clin Oncol 2014;32:Abstracte17653.

125. Kris MG, Gucalp A, Epstein AS, et al. Assessing the performanceof Watson for oncology, a decision support system, using actualcontemporary clinical cases. In 2015 ASCO Annual Meeting,Edition. J Clin Oncol 2015;33:Abstract 8023.

126. Gray SW, Hicks-Courant K, Cronin A, et al. Physicians’ attitudesabout multiplex tumor genomic testing. J Clin Oncol 2014;32:1317–23.

127. Aronson SJ, Rehm HL. Building the foundation for genomics inprecision medicine. Nature 2015;526:336–42.

128. Richards S, Aziz N, Bale S, et al. Standards and guidelines for theinterpretation of sequence variants: a joint consensus recommendationof the American College of Medical Genetics and Genomics and theAssociation for Molecular Pathology.Genet Med 2015;17:405–24.

129. Mendel G. Versuche über Plflanzenhybriden. Verhandlungen desnaturforschenden Vereines in Brünn, 1866;4:3–47.

Horak P, et al. ESMO Open 2016;1:e000094. doi:10.1136/esmoopen-2016-000094 11

Open Access

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from

pitfallsclinical oncology: strategies, promises and Integrating next-generation sequencing into

Peter Horak, Stefan Fröhling and Hanno Glimm

doi: 10.1136/esmoopen-2016-0000942016 1: ESMO Open

http://esmoopen.bmj.com/content/1/5/e000094Updated information and services can be found at:

These include:

References http://esmoopen.bmj.com/content/1/5/e000094#ref-list-1

This article cites 123 articles, 23 of which you can access for free at:

Open Access

http://creativecommons.org/licenses/by-nc/4.0/non-commercial. See: provided the original work is properly cited and the use isnon-commercially, and license their derivative works on different terms, permits others to distribute, remix, adapt, build upon this workCommons Attribution Non Commercial (CC BY-NC 4.0) license, which This is an Open Access article distributed in accordance with the Creative

serviceEmail alerting

box at the top right corner of the online article. Receive free email alerts when new articles cite this article. Sign up in the

CollectionsTopic Articles on similar topics can be found in the following collections

(225)Open access

Notes

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://group.bmj.com/subscribe/To subscribe to BMJ go to:

group.bmj.com on May 18, 2018 - Published by http://esmoopen.bmj.com/Downloaded from