Understanding the molecular nature of cancer is key to administering the most effective treatments. Cancer patients are therefore often subject to invasive tissue biopsies to discern the dominant traits of a tumour. However this approach does not necessarily capture the whole story, with rapid changes in the cancer genome known to occur as the disease progresses, in response to treatment, and during metastasis, often in a small proportion of cells. Prostate cancer in particular is known to recur in 20-30 percent of men after five years, in spite of initial curative treatment. Furthermore, metastatic prostate cancer often spreads to the bone, making tissue biopsies technically challenging and therefore limiting repeated sampling over the course of the disease. With this in mind, Michael Speicher and Ellen Heitzer from the Medical University of Graz, Austria, and colleagues, developed a ‘liquid biopsy’ method for prostate cancer patients, as published in their study in Genome Medicine, which this year received BioMed Central’s 8th Annual Research Award. Here winning author Heitzer, alongside Speicher, discuss what is needed for this approach to become routine clinical practice, and what the future of non-invasive cancer diagnostics entails.
How did your interest in genomic alterations and their impact on disease susceptibility come about, particularly with regards to cancer?
Despite the massive technological advances over the past half century, cancer is still a leading cause of death worldwide. Not many people have been spared the loss of a friend or loved one to cancer. It is not just one disease, it is many diseases and the more cancer genomes we decipher, the more molecular changes we discover. Most cancer cells harbour numerous genetic changes and even cells within a tumour can be completely different. It is quite astonishing that cancer cells are able to become more invasive and deadly as they accumulate more genetic changes, rather than dying because of the resulting chromosomal chaos.
Based on a deeper understanding of tumours at the molecular level, the ‘one size fits all’ approach to cancer treatment has dramatically changed and a new generation of so called ‘targeted therapies’ is available today. Unfortunately, with the implementation of these therapies into clinical practice, new challenges arose. The major problems are tumour heterogeneity and the lack of genetic follow up data. We therefore wanted to develop solutions to improve therapy management for cancer patients.
What led to your study in Genome Medicine, and what was the main goal of the research?
To tailor anti-cancer treatments as much as possible guiding biomarkers are needed. While prognostic markers help to identify individuals who are at high risk of recurrence of their cancer and should therefore receive further adjuvant therapy, predictive biomarkers help to identify therapy targets and subgroups of patients who are most likely to benefit from a given therapy. However, even the best marker is useless if it is not accessible or cannot be repeatedly evaluated.
Therapy decisions are mainly based on biopsies and the treatment is designed to target the dominant clones. Thus, subclones that are also relevant for the treatment or already harbour primary resistance mechanisms might be missed by biopsies. In addition cancer cells accumulate new genetic changes as a consequence of tumour progression and the selective pressure of cancer therapies. Furthermore, metastases often show different changes than the primary tumour and in most cases we are not able to monitor these changes with simple biopsies. In general progression is only noted if it is already clinically obvious after evaluation with imaging techniques and clinical tumour markers that are not accurate enough.
One possibility to overcome these limitations is the establishment of biomarkers from easily accessible biofluids like blood or plasma, the so called liquid biopsy. The analysis of tumour-specific changes on circulating tumour cells (CTC) and cell-free circulating tumour DNA (ctDNA) is beneficial when compared to tissue biopsies as repeated sampling is easily achievable. Furthermore, ctDNA reflects genetic changes including point mutations, methylation patterns, copy number aberrations and structural rearrangements from different tumour locations, and can therefore be used as surrogate markers for the entire tumour genome.
We therefore aimed to develop a sequencing based technique where it is possible to monitor tumour genomes non-invasively from plasma on both a genome-wide and a gene-specific level. The main goal, however, was to establish a method that can deliver relevant information in a very short time and at reasonable costs, such that this method can eventually be implemented in routine clinical practice.
What did you find and why was this exciting/interesting?
Tumours release cell-free DNA in to the circulation, which is then referred to as circulating tumour DNA (ctDNA). As ctDNA reflects genetic changes from different tumour locations it is more representative of the mutational heterogeneity in a tumour than a tissue biopsy. We developed a sequencing based approach called plasma-Seq where it is possible to reconstruct tumour genomes non-invasively from plasma on both a genome-wide and a gene-specific level. We were able to establish copy number variations within 24 hours and with the addition of a targeted approach we can also detect specific driver mutations that might be used as therapy targets or for monitoring purposes.
The big advantage of our approach compared to other available monitoring tools is that we employ a benchtop sequencer, the Illumina MiSeq that is affordable for smaller labs and generates sequence data in a very short time and at reasonable costs. It is most important for patients as well as clinicians to obtain data for therapy decisions in a very short time frame to enable a rapid response to any relevant genetic changes.
What are the clinical implications of your findings?
The most important clinical implication is that we are able to systematically track the genomic evolution of cancer non-invasively in a very cost-effective manner. Knowing early if resistance has developed would allow patients to switch therapies before the progression is clinically obvious. This might therefore contribute to improved survival of cancer patients. Furthermore, anti-cancer treatments are increasingly administered according to ‘druggable targets’ rather than the tumour entity. There is no question that continuous monitoring of the evolution of tumour genomes is inevitable. As tissue biopsies are invasive and some tumours are inaccessible, liquid biopsy represents a minimal invasive method where repeated sampling is easily achievable.
What barriers must be overcome to translate your ‘liquid biopsy’ approach to routine clinical practice?
The liquid biopsy is a very promising approach, however there are some issues that need to be solved before it can be actually implemented into clinical practice. First, the biology of the release of tumour DNA is not fully elucidated yet. Although plenty of studies indicate that ctDNA is correlated with tumour burden, tumour-specific changes cannot be identified in all tumour patients, not even in all patients showing metastasis. Second, the amount of ctDNA in the circulation varies dramatically and can range from less than one percent to more than 90 percent. Third, the mechanism of DNA clearance from plasma is also poorly understood and it is not known how other factors such as circadian rhythms, inflammation or particular therapies influence release and clearance mechanisms. Finally, the predictive and prognostic value of our approach, and ctDNA in general, needs to be evaluated in large clinical studies.
It is absolutely critical to establish standard operating procedures at both pre-analytical and analytical levels. To overcome logistical challenges that might represent hurdles for clinical implementation large-scale prospective studies are necessary in order to standardise our technique and to assess reproducibility. A further caveat is that for many cancer and/or tumour subtypes predictive markers are lacking and many mechanisms of progression or acquired resistance are not yet understood. In addition, in scenarios where only low levels of ctDNA are present (e.g. early stage or minimal residual cancer) our approach might not be applicable as its sensitivity is limited.
What impact do you think continued advances in next-generation sequencing will have on cancer diagnosis and therapy?
There is no doubt that next-generation sequencing (NGS) techniques have already and will continue to revolutionise cancer management. In terms of non-invasive cancer diagnostics there are basically two NGS-based approaches, one that is targeted and one that is genome-wide.
The targeted approach involves analysing known genetic changes in a primary tumour from a small set of frequently occurring driver mutations (e.g. mutations in KRAS, EGFR, etc), with implications for therapy decisions. A recent study from Bettegowda and colleagues (Sci Transl Med. 2014, Feb 19, 6, 224:224ra24) analysed a large set of cancer patients with different tumour entities and tumour stages. The sensitivity of ctDNA for detection of clinically relevant KRAS gene mutations was 87.2 percent and its specificity was 99.2 percent. However, they used highly sensitive methods including digital PCR and the Safe-SeqS method that was previously established by the same group (Proc Natl Acad Sci U S A. 2011, Jun 7, 108, 23:9530-5). This approach represents a good method to detect tumour specific mutations even at very low levels and allows a clear distinction from the background. Furthermore, they screened for translocations that are highly tumour-specific and can be used to detect tumour-specific changes at very low levels or for the identification of minimal residual disease. In our study we were also able to identify structural rearrangements after targeted enrichment of a chromosomal region that is frequently involved in translocations. Another technique to track down tumour-specific mutations in plasma was developed by Forshew and colleagues (Sci Transl Med. 2012, May 30, 4, 136:136ra68), called tagged-amplicon deep sequencing (TAm-Seq).
The genome-wide approach involves identifying de novo tumour-derived chromosomal alterations through massively parallel direct sequencing of DNA from the plasma, akin to our plasma-Seq method. This means sequencing the primary tumour is not necessary. Analysis of the ctDNA could deliver sufficient information for therapy management. Furthermore, such approaches are applicable to all patients as they do not rely on recurrent genetic changes. Dennis Lo and colleagues in Hong Kong, China, were among the first that established genome-wide profiles from plasma and further developed this technique by a combined assessment of hypomethylation and cancer-associated copy number aberrations. Leary and colleagues (Sci Transl Med. 2010, Feb 24, 2, 20:20ra14) also developed a method called PARE (personalised analysis of rearranged ends), to identify translocations in solid tumours and applied this approach in plasma DNA samples, where they identified several chromosomal copy number changes and rearrangements, including amplification of cancer driver genes such as ERBB2 and CDK6. Murtaza and colleagues (Nature. 2013, May 2, 497, 7447:108-12) performed exome sequencing of plasma DNA samples and followed multiple courses of treatment. Quantification of allele fractions in plasma identified an increased representation of mutant alleles in association with the emergence of therapy resistance.
There are therefore numerous approaches to analyse tumours non-invasively, but it is not clear yet which method will emerge as the best one. In contrast to our method where it is possible to obtain clinically relevant information in a cost-effective and fast manner – as demonstrated in our recent PLOS Genetics paper (PLoS Genet. 2014, Mar 27, 10, 3:e1004271) – most other techniques are still too expensive and time consuming. However, sequencing costs will further drop and this field of technology is continuously evolving. It is just a matter of time until technical advances and cost reductions will allow the implementation of genome-wide approaches with high resolution as a routine tool in laboratory medicine. Meanwhile, clinical standards are needed in order to compare and validate methods, also taking into account different diagnostic centres and settings.
Why did you choose to publish your findings in an Open Access journal?
The main reason for publishing in an Open Access (OA) journal was so that any researcher can read and build on our findings. As the research findings are publically available, an individual researcher may also gain more visibility, recognition, readership, and citations than in traditional journals. Although impact factor is still one of the principal metrics of an article and a measure of the reputation of a researcher, in our opinion it does not necessarily reflect the value of a scientific work. It is rather the readership that judges the relevance of an article which can be, for example, nicely reflected by an Altmetric score that provides information on the usage and dissemination of published article. It should however be noted that there are many OA journals that are at the top of their disciplinary categories and several studies indicate that there is no dramatic difference in citations compared to traditional journals.
What’s next for your research?
We are currently focusing on the three big Cs: prostate, colorectal, and breast cancer. Regarding colorectal cancer we recently published a study in PLOS Genetics, in which we used the plasma-Seq approach to assess the genetic evolution of tumours under anti-EGFR therapy. We observed that specific copy number changes of genes, such as KRAS, MET, or ERBB2, can be acquired under therapy and determine responsiveness to therapy. In the near future we aim to validate our approach in controlled, prospective clinical studies, and in doing so take one step closer to implementation of liquid biopsy in routine clinical practice.
We are also trying to better understand tumour progression and resistance mechanisms. Whilst establishing genome-wide copy number changes, we have frequently observed numerous focal amplifications that might include genes that are involved in tumour progression. We therefore may be able to identify new treatable targets or driver genes.
In addition, we are working on more sensitive methods to be able to monitor tumour evolution in early stages of cancer and in those patients in which the amount of tumour DNA is currently below the detection limit of our established assay.